framework,version,device,op_name,kernel_source,moe_dtype,num_tokens,hidden_size,inter_size,topk,num_experts,moe_tp_size,moe_ep_size,distribution,latency
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,1,balanced,0.21264533201853433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,1,balanced,0.3091626763343811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,1,balanced,0.49208001295725506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,1,balanced,0.4984480142593384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,1,balanced,0.6031733353932699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,1,balanced,0.6058880090713501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,1,balanced,0.6079626480738322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,1,balanced,0.6116000016530355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,1,balanced,0.6189493338267008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,1,balanced,0.6197706858317057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,1,balanced,0.6277173360188802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,1,balanced,0.6380000114440918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,1,balanced,0.6454079945882162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,1,balanced,0.6588106552759806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,1,balanced,0.8480853239695231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,1,balanced,0.8592267036437988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,1,balanced,0.971946636835734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,1,balanced,1.3495945930480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,1,balanced,1.6610719362894695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,1,balanced,2.6012852986653647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,1,balanced,3.2148640950520835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,1,balanced,5.241429328918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,1,balanced,7.348314921061198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,1,balanced,0.1695680022239685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,1,balanced,0.22126400470733643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,1,balanced,0.33530131975809735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,1,balanced,0.34388267993927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,1,balanced,0.41709331671396893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,1,balanced,0.4266666571299235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,1,balanced,0.42904531955718994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,1,balanced,0.440287987391154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,1,balanced,0.4463733434677124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,1,balanced,0.4498080015182495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,1,balanced,0.4769546588261922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,1,balanced,0.48330668608347577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,1,balanced,0.5099626779556274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,1,balanced,0.5470933516820272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,1,balanced,0.835482676823934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,1,balanced,0.8655786514282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,1,balanced,0.894005298614502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,1,balanced,12.814992268880209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,1,balanced,1.2515947024027507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,1,balanced,1.5722613334655762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,1,power_law_1.01,0.2095103979110718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,1,power_law_1.01,0.2681152105331421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,1,power_law_1.01,0.3127039909362793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,1,balanced,2.3263039588928223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,1,power_law_1.01,0.479417610168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,1,power_law_1.01,0.6022592067718506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,1,balanced,2.992725372314453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,1,power_law_1.01,0.591264009475708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,1,power_law_1.01,0.6081471920013428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,1,power_law_1.01,0.6116479873657227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,1,balanced,4.481333414713542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,1,power_law_1.01,0.6183551788330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,1,power_law_1.01,0.6167424201965332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,1,power_law_1.01,0.6272640228271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,1,balanced,17.905946095784504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,1,power_law_1.01,0.6691711902618408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,1,power_law_1.01,0.16550400257110595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,1,balanced,5.891909281412761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,1,power_law_1.01,0.2089792013168335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,1,power_law_1.01,0.6994112014770508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,1,power_law_1.01,0.2234112024307251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,1,power_law_1.01,0.7807040214538574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,1,power_law_1.01,0.847116756439209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,1,power_law_1.01,1.0102463722229005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,1,balanced,8.85475222269694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,1,power_law_1.01,0.32403199672698973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,1,power_law_1.01,0.41816320419311526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,1,power_law_1.01,0.4219520092010498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,1,power_law_1.01,0.42997121810913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,1,power_law_1.01,1.1497983932495117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,1,balanced,0.060362666845321655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,1,balanced,0.0857973297437032
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,1,balanced,0.1399733324845632
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,1,balanced,0.140773336092631
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,1,balanced,0.1439306636651357
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,1,balanced,0.1453546682993571
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,1,balanced,0.17988266547520956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,1,balanced,0.18177600701649985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,1,power_law_1.01,1.5168512344360352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,1,balanced,0.18028799692789713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,1,balanced,0.18345065911610922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,1,balanced,0.1848213275273641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,1,balanced,12.493403116861979
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,1,balanced,0.19458667437235513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,1,balanced,0.19738666216532388
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,1,balanced,0.20207999149958292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,1,balanced,0.3652213414510091
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,1,balanced,0.29707199335098267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,1,balanced,0.2283786733945211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,1,power_law_1.01,1.8726720809936523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,1,balanced,0.601418654123942
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,1,power_law_1.01,0.0580672025680542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,1,balanced,0.2696746587753296
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,1,power_law_1.01,0.07216640114784241
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,1,balanced,0.3679573138554891
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,1,power_law_1.01,0.09711359739303589
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,1,balanced,0.4556479851404826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,1,power_law_1.01,0.13636480569839476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,1,balanced,0.6495519876480103
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,1,power_law_1.01,0.14266239404678344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,1,balanced,0.8364746570587158
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,1,power_law_1.01,0.1593791961669922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,1,power_law_1.01,2.595270347595215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,1,power_law_1.01,0.18021759986877442
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,1,balanced,1.2423840363820393
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,1,power_law_1.01,0.19912960529327392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,1,power_law_1.01,0.1759168028831482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,1,balanced,1.6395039558410645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,1,power_law_1.01,0.18584959506988524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,1,balanced,28.3660151163737
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,1,power_law_1.01,0.20484480857849122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,1,balanced,2.4424266815185547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,1,power_law_1.01,0.19459199905395508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,1,power_law_1.01,0.20555520057678223
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,1,power_law_1.01,0.23430399894714354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,1,power_law_1.01,3.31383056640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,1,power_law_1.01,0.4372992038726807
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,1,power_law_1.01,0.22741758823394775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,1,balanced,3.425914764404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,1,power_law_1.01,0.4600128173828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,1,power_law_1.01,0.4755263805389404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,1,power_law_1.01,0.5235136032104493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,1,balanced,22.75927988688151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,1,power_law_1.01,0.5679872035980225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,1,power_law_1.01,0.6244416236877441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,1,power_law_1.01,5.136358261108398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,1,power_law_1.01,0.7230271816253662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,1,power_law_1.01,0.7955327987670898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,1,power_law_1.01,0.23704960346221923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,1,power_law_1.01,0.9107711791992188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,1,power_law_1.01,0.26817920207977297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,1,power_law_1.01,1.0598591804504394
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,1,power_law_1.01,0.3495680093765259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,1,power_law_1.01,1.4086015701293946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,1,balanced,0.1425226628780365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,1,balanced,0.20054399967193604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,1,balanced,0.30957865715026855
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,1,power_law_1.01,0.36820480823516843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,1,balanced,0.31390400727589923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,1,balanced,0.31441599130630493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,1,balanced,0.31804267565409344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,1,balanced,0.32124267021814984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,1,balanced,0.32389867305755615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,1,balanced,40.480010986328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,1,power_law_1.01,1.74453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,1,balanced,33.93939208984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,1,power_law_1.01,0.4352128028869629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,1,power_law_1.2,0.2120896100997925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,1,balanced,0.33086933692296344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,1,balanced,0.33725865681966144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,1,power_law_1.2,0.2746880054473877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,1,balanced,0.34279998143513996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,1,power_law_1.2,0.3400063991546631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,1,power_law_1.2,0.49791998863220216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,1,power_law_1.2,0.5892479896545411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,1,power_law_1.2,0.604640007019043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,1,power_law_1.2,0.16858240365982055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,1,power_law_1.2,0.6090432167053222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,1,power_law_1.01,2.4651071548461916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,1,power_law_1.01,0.140774405002594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,1,power_law_1.01,0.5503744125366211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,1,power_law_1.2,0.6104576110839843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,1,power_law_1.01,0.17855360507965087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,1,balanced,0.35507198174794513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,1,power_law_1.2,0.6170752048492432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,1,balanced,0.3623093366622925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,1,balanced,0.40672000249226886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,1,balanced,0.5624053478240967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,1,power_law_1.2,0.6311039924621582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,1,balanced,0.5889653364817301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,1,power_law_1.2,0.6425216197967529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,1,balanced,0.6312106847763062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,1,balanced,0.8728853066762289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,1,power_law_1.2,0.6810751914978027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,1,power_law_1.2,0.20062079429626464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,1,balanced,1.100933313369751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,1,power_law_1.2,0.2525631904602051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,1,power_law_1.2,0.6941567897796631
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,1,power_law_1.01,0.7273536205291748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,1,power_law_1.01,0.21336960792541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,1,power_law_1.2,0.3307136058807373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,1,power_law_1.2,0.7584320068359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,1,power_law_1.01,0.3054656028747559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,1,power_law_1.2,0.409881591796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,1,power_law_1.01,0.30741760730743406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,1,power_law_1.2,0.4230016231536865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,1,power_law_1.01,0.3167167901992798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,1,power_law_1.2,0.4294591903686523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,1,power_law_1.2,0.8328319549560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,1,power_law_1.2,0.4406144142150879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,1,power_law_1.2,0.45969281196594236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,1,power_law_1.2,0.4798719882965088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,1,balanced,1.5953067143758137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,1,power_law_1.2,0.53056640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,1,power_law_1.2,0.5761663913726807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,1,balanced,2.024831930796305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,1,power_law_1.2,1.0814592361450195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,1,power_law_1.01,7.344127655029297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,1,power_law_1.01,0.3127360105514526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,1,power_law_1.01,0.3241919994354248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,1,balanced,3.031813303629557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,1,power_law_1.01,0.331443190574646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,1,power_law_1.01,3.2231937408447267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,1,power_law_1.01,0.350598406791687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,1,power_law_1.01,0.39375360012054444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,1,power_law_1.2,1.1748031616210937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,1,balanced,3.9297707875569663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,1,power_law_1.01,0.41829118728637693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,1,power_law_1.01,0.45331201553344724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,1,power_law_1.2,0.6273407936096191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,1,power_law_1.01,0.49626879692077636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,1,power_law_1.2,1.5032320022583008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,1,power_law_1.01,0.5470975875854492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,1,power_law_1.01,0.9189056396484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,1,balanced,5.866298675537109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,1,power_law_1.01,0.5962751865386963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,1,power_law_1.2,0.736633586883545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,1,power_law_1.2,0.7893119812011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,1,power_law_1.2,1.8643903732299805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,1,power_law_1.01,0.7090559959411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,1,power_law_1.2,0.89169921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,1,balanced,7.664661407470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,1,power_law_1.01,0.9467840194702148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,1,power_law_1.2,1.0527680397033692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,1,power_law_1.01,4.660095977783203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,1,power_law_1.01,11.464249420166016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,1,power_law_1.01,1.160268783569336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,1,balanced,11.532085418701172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,1,power_law_1.2,2.578950309753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,1,power_law_1.2,1.435878372192383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,1,power_law_1.01,1.637727928161621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,1,balanced,15.508837381998697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,1,power_law_1.2,1.7573951721191405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,1,power_law_1.01,2.136025619506836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,1,power_law_1.01,6.0793407440185545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,1,power_law_1.2,2.4883647918701173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,1,power_law_1.2,3.325503921508789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,1,power_law_1.2,0.1428928017616272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,1,power_law_1.2,0.17003519535064698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,1,power_law_1.2,0.20133121013641359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,1,power_law_1.2,0.3128319978713989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,1,power_law_1.2,0.30820479393005373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,1,power_law_1.2,0.3118272066116333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,1,power_law_1.2,0.32181758880615235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,1,power_law_1.2,0.3252351999282837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,1,power_law_1.2,0.3371840000152588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,1,power_law_1.2,0.33902080059051515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,1,power_law_1.01,1.3020607948303222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,1,power_law_1.01,3.0462528228759767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,1,power_law_1.2,0.3625727891921997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,1,power_law_1.2,0.4297920227050781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,1,power_law_1.2,0.45374717712402346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,1,power_law_1.2,3.215296173095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,1,power_law_1.2,0.5000448226928711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,1,power_law_1.2,0.544159984588623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,1,power_law_1.2,5.192287826538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,1,power_law_1.01,4.024179077148437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,1,power_law_1.2,0.5851776123046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,1,power_law_1.2,0.7261760234832764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,1,power_law_1.2,0.9356736183166504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,1,power_law_1.2,1.1416000366210937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,1,power_law_1.2,4.682137680053711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,1,power_law_1.01,15.925010681152344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,1,power_law_1.2,1.6276800155639648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,1,power_law_1.2,0.05863040089607239
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,1,power_law_1.2,0.06972159743309021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,1,power_law_1.2,0.09275519847869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,1,power_law_1.2,0.14017280340194702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,1,power_law_1.2,0.1415295958518982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,1,power_law_1.2,6.132249450683593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,1,power_law_1.2,0.15943039655685426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,1,power_law_1.2,0.18046079874038695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,1,power_law_1.2,0.20559360980987548
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,1,power_law_1.2,0.18246400356292725
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,1,power_law_1.2,0.18888959884643555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,1,power_law_1.2,0.20650880336761473
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,1,power_law_1.2,0.1933184027671814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,1,power_law_1.2,0.2060159921646118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,1,power_law_1.01,8.989068603515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,1,power_law_1.2,0.236844801902771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,1,power_law_1.01,5.924224090576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,1,power_law_1.2,7.192972564697266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,1,power_law_1.2,2.1290496826171874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,1,power_law_1.01,1.6931135177612304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,1,power_law_1.2,0.24936959743499756
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,1,power_law_1.2,0.2484287977218628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,1,power_law_1.2,0.26673920154571534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,1,power_law_1.01,12.023174285888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,1,power_law_1.2,0.28095359802246095
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,1,power_law_1.2,0.3241663932800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,1,power_law_1.2,0.46340479850769045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,1,power_law_1.2,3.050150489807129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,1,power_law_1.2,0.6799232006072998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,1,power_law_1.2,0.712608003616333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,1,power_law_1.2,9.096300506591797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,1,power_law_1.01,26.027182006835936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,1,power_law_1.2,4.0451007843017575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,1,power_law_1.01,7.825689697265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,1,power_law_1.01,2.471353530883789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,1,power_law_1.2,0.892147159576416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,1,power_law_1.2,11.768505859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,1,power_law_1.01,20.028562927246092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,1,power_law_1.01,11.622783660888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,1,power_law_1.2,1.394547176361084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,1,power_law_1.2,5.926758575439453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,1,power_law_1.2,16.153497314453126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,1,power_law_1.2,12.277516937255859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,1,power_law_1.01,15.3905029296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,1,balanced,0.31224532922108966
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,1,power_law_1.01,3.521772766113281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,1,power_law_1.2,1.6718784332275392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,1,power_law_1.01,27.785598754882812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,1,power_law_1.01,36.51182861328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,1,balanced,0.4739253520965576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,1,balanced,0.8343786398569742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,1,balanced,0.22593067089716592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,1,balanced,0.31653332710266113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,1,balanced,0.5918826659520467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,1,balanced,0.5924746592839559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,1,balanced,0.5863306522369385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,1,balanced,0.5966613292694092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,1,balanced,0.6051679849624634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,1,balanced,0.6239466667175293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,1,balanced,0.6483146746953329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,1,balanced,0.6455733378728231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,1,balanced,0.6741173267364502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,1,balanced,0.7153813044230143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,1,balanced,0.7517759799957275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,1,balanced,0.8187946478525797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,1,balanced,1.2317919731140137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,1,balanced,1.2777973016103108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,1,balanced,1.388879934946696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,1,balanced,0.8386507034301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,1,balanced,2.000666618347168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,1,balanced,2.5541332562764487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,1,balanced,3.786975860595703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,1,balanced,4.906336148579915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,1,power_law_1.2,7.838918304443359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,1,balanced,7.360650380452474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,1,balanced,9.686506907145182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,1,balanced,0.9345013300577799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,1,balanced,0.9392639795939127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,1,balanced,0.937824010848999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,1,balanced,0.9401013056437174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,1,balanced,0.9497066338857015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,1,balanced,0.9545706907908121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,1,balanced,0.9607146581013998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,1,balanced,0.9810346762339274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,1,balanced,0.9860586325327555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,1,balanced,1.0185173352559407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,1,balanced,14.813765207926432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,1,balanced,1.491605281829834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,1,power_law_1.2,26.030215454101562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,1,balanced,1.5502452850341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,1,balanced,1.5749492645263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,1,balanced,0.07728533446788788
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,1,balanced,0.1267466644446055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,1,balanced,2.238464037577311
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,1,balanced,0.2188426653544108
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,1,balanced,0.22001065810521445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,1,balanced,0.2221440076828003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,1,balanced,0.22339733441670737
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,1,balanced,0.3545866807301839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,1,balanced,2.8736747105916343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,1,balanced,0.35930665334065753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,1,balanced,0.35898133118947345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,1,balanced,0.3612746795018514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,1,balanced,0.1869600017865499
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,1,balanced,0.36208534240722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,1,balanced,0.29340799649556476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,1,balanced,0.37789865334828693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,1,balanced,0.5384586652119955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,1,balanced,0.379253347714742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,1,balanced,0.38554131984710693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,1,balanced,4.104954719543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,1,balanced,0.528005321820577
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,1,balanced,0.46325866381327313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,1,balanced,0.42440001169840497
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,1,balanced,0.7944053014119467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,1,balanced,0.45138665040334064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,1,balanced,0.6236746708552042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,1,balanced,5.6993865966796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,1,balanced,0.827018658320109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,1,balanced,22.57140350341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,1,balanced,1.2047093709309895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,1,balanced,0.5408213138580322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,1,balanced,0.5431626637776693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,1,balanced,0.5462986628214518
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,1,balanced,1.6065492630004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,1,balanced,0.5534826517105103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,1,balanced,0.5577813386917114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,1,balanced,0.5666613181432089
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,1,balanced,2.397450606028239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,1,balanced,0.5690933465957642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,1,balanced,0.5785653193791708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,1,balanced,0.5929386615753174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,1,balanced,9.497088114420572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,1,balanced,0.6039573351542155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,1,balanced,3.4160852432250977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,1,balanced,0.6547733147939047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,1,balanced,0.8686506748199463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,1,balanced,0.8904000123341879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,1,balanced,0.9351840019226074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,1,balanced,1.2914293607076008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,1,balanced,5.425914764404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,1,balanced,1.6878239313761394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,1,balanced,2.427173296610514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,1,balanced,13.217466990152994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,1,balanced,6.457706451416016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,1,balanced,3.1765867869059243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,1,power_law_1.01,0.3311295986175537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,1,power_law_1.01,0.6766784191131592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,1,power_law_1.01,0.7128640174865722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,1,balanced,4.675818761189778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,1,power_law_1.01,0.8768128395080567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,1,power_law_1.01,0.9177663803100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,1,power_law_1.01,0.9392255783081055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,1,power_law_1.01,0.9354623794555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,1,power_law_1.01,0.9750271797180176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,1,balanced,38.305920918782554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,1,power_law_1.01,0.9960063934326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,1,power_law_1.01,1.0206720352172851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,1,power_law_1.01,1.0407296180725099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,1,balanced,6.122805277506511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,1,power_law_1.2,20.65809326171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,1,balanced,21.043163299560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,1,power_law_1.01,1.234995174407959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,1,power_law_1.01,1.296076774597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,1,power_law_1.2,2.452934455871582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,1,power_law_1.01,1.3971327781677245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,1,balanced,9.215930938720703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,1,power_law_1.01,1.5182080268859863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,1,power_law_1.01,1.6665536880493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,1,power_law_1.01,1.9040063858032226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,1,balanced,12.030043284098307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,1,power_law_1.01,2.4932479858398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,1,balanced,30.411348978678387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,1,power_law_1.01,3.221478271484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,1,balanced,18.71176528930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,1,power_law_1.01,4.66253433227539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,1,balanced,58.9687245686849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,1,power_law_1.01,0.22417919635772704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,1,power_law_1.01,0.28010239601135256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,1,power_law_1.01,0.36369919776916504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,1,power_law_1.01,0.5541888236999511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,1,power_law_1.01,0.5758656024932861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,1,power_law_1.01,0.5980607986450195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,1,power_law_1.01,0.6151616096496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,1,power_law_1.01,0.6426432132720947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,1,power_law_1.01,0.6942783832550049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,1,balanced,24.639668782552082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,1,power_law_1.2,11.63751678466797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,1,power_law_1.01,6.052793502807617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,1,power_law_1.01,0.1856384038925171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,1,power_law_1.01,0.25391359329223634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,1,power_law_1.01,0.2758016109466553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,1,power_law_1.01,0.5186816215515136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,1,power_law_1.01,0.7663424015045166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,1,power_law_1.01,0.5339072227478028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,1,power_law_1.01,0.8246912002563477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,1,power_law_1.01,0.5434112071990966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,1,power_law_1.01,0.9495360374450683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,1,power_law_1.01,0.5410687923431396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,1,balanced,48.06500752766927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,1,power_law_1.01,0.5475776195526123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,1,power_law_1.01,0.9835136413574219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,1,power_law_1.01,0.5665919780731201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,1,power_law_1.01,1.0980863571166992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,1,power_law_1.01,0.5715007781982422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,1,power_law_1.01,0.5910336017608643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,1,power_law_1.01,1.2217856407165528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,1,power_law_1.01,0.628876781463623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,1,power_law_1.01,9.175007629394532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,1,power_law_1.01,0.6509056091308594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,1,power_law_1.01,0.7471744060516358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,1,power_law_1.01,0.8404864311218262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,1,power_law_1.01,1.393772792816162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,1,power_law_1.01,1.6554752349853517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,1,power_law_1.01,0.9107199668884277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,1,power_law_1.01,2.274835205078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,1,power_law_1.01,12.70579833984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,1,power_law_1.01,1.0922368049621582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,1,power_law_1.01,1.4348031997680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,1,power_law_1.01,2.8406143188476562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,1,power_law_1.01,1.826585578918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,1,power_law_1.01,4.029663848876953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,1,power_law_1.01,2.5174528121948243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,1,balanced,68.01303100585938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,1,power_law_1.2,37.53489379882812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,1,power_law_1.01,5.220940780639649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,1,power_law_1.2,0.31370880603790285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,1,power_law_1.2,0.4011712074279785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,1,power_law_1.2,0.522822380065918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,1,power_law_1.2,0.8415360450744629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,1,power_law_1.2,0.8968319892883301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,1,power_law_1.01,3.3047039031982424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,1,power_law_1.2,0.929593563079834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,1,power_law_1.2,0.9225024223327637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,1,power_law_1.2,0.9713343620300293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,1,power_law_1.2,1.0011263847351075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,1,power_law_1.2,0.9606399536132812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,1,power_law_1.2,1.0397567749023438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,1,power_law_1.01,4.793644714355469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,1,power_law_1.2,1.2504768371582031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,1,power_law_1.2,1.2929856300354003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,1,power_law_1.01,7.567987060546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,1,power_law_1.2,1.3902463912963867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,1,power_law_1.2,1.4740415573120118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,1,power_law_1.01,6.245196914672851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,1,power_law_1.01,19.827098083496093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,1,power_law_1.2,1.6331968307495117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,1,power_law_1.2,2.0080575942993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,1,power_law_1.2,2.521900749206543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,1,power_law_1.2,3.191609573364258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,1,power_law_1.01,9.364096069335938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,1,power_law_1.2,3.3915008544921874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,1,power_law_1.01,9.961241912841796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,1,power_law_1.2,4.372646331787109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,1,power_law_1.2,0.24062719345092773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,1,power_law_1.2,0.2903104066848755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,1,power_law_1.2,0.3822144031524658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,1,power_law_1.01,28.76909484863281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,1,power_law_1.01,0.07670400142669678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,1,power_law_1.2,0.5950975894927979
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,1,power_law_1.01,0.11664639711380005
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,1,power_law_1.01,0.14202879667282103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,1,power_law_1.2,0.578380823135376
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,1,power_law_1.01,0.21802880764007568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,1,power_law_1.2,0.5948736190795898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,1,power_law_1.01,0.23255679607391358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,1,power_law_1.2,29.738311767578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,1,power_law_1.2,0.6293632030487061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,1,power_law_1.2,0.647814416885376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,1,power_law_1.2,0.7421823978424072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,1,power_law_1.2,0.7767295837402344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,1,power_law_1.2,0.8644288063049317
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,1,power_law_1.01,0.2742399930953979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,1,power_law_1.2,0.9259584426879883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,1,power_law_1.01,0.34627840518951414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,1,power_law_1.01,0.41432957649230956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,1,power_law_1.2,0.9896703720092773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,1,power_law_1.2,6.181695938110352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,1,power_law_1.01,0.32269439697265623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,1,power_law_1.01,0.3608896017074585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,1,power_law_1.2,1.120633602142334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,1,power_law_1.01,0.4256703853607178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,1,power_law_1.01,0.34514560699462893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,1,power_law_1.2,1.225171184539795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,1,power_law_1.01,0.37411839962005616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,1,power_law_1.2,1.3820672035217285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,1,power_law_1.01,0.45442562103271483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,1,power_law_1.01,12.278336334228516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,1,power_law_1.2,1.639379119873047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,1,power_law_1.01,0.392793607711792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,1,power_law_1.2,2.2259328842163084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,1,power_law_1.01,0.43924479484558104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,1,power_law_1.01,0.507859182357788
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,1,power_law_1.01,0.508454418182373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,1,power_law_1.2,2.821932792663574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,1,power_law_1.01,0.5810688018798829
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,1,power_law_1.01,0.7681344032287598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,1,power_law_1.2,9.368883514404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,1,power_law_1.01,0.950540828704834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,1,power_law_1.2,4.031155014038086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,1,power_law_1.01,15.003929138183594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,1,power_law_1.2,5.196998214721679
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,1,power_law_1.01,1.373971176147461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,1,power_law_1.2,12.346495819091796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,1,power_law_1.2,7.598297882080078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,1,power_law_1.01,1.7319168090820312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,1,power_law_1.2,0.1854143977165222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,1,power_law_1.2,0.25335040092468264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,1,power_law_1.2,0.29421439170837405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,1,power_law_1.2,0.5037055969238281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,1,power_law_1.2,0.5389632225036621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,1,power_law_1.2,0.5436351776123047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,1,power_law_1.2,0.5482240200042725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,1,power_law_1.2,0.5531263828277588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,1,power_law_1.2,0.571065616607666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,1,power_law_1.2,0.5748672008514404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,1,power_law_1.2,0.5998400211334228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,1,power_law_1.2,0.6335423946380615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,1,power_law_1.2,0.6511936187744141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,1,power_law_1.2,0.7741568088531494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,1,power_law_1.2,0.8445887565612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,1,power_law_1.2,0.9204416275024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,1,power_law_1.2,1.1069439888000487
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,1,power_law_1.01,2.55861759185791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,1,power_law_1.2,1.4285056114196777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,1,power_law_1.2,9.99649887084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,1,power_law_1.2,1.79354248046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,1,power_law_1.2,15.483392333984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,1,power_law_1.2,2.539616012573242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,1,power_law_1.01,18.32981719970703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,1,power_law_1.01,21.038777160644532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,1,power_law_1.2,3.275603103637695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,1,power_law_1.01,3.358540725708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,1,power_law_1.2,14.905389404296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,1,power_law_1.2,4.79552001953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,1,power_law_1.01,46.90911865234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,1,power_law_1.2,20.721343994140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,1,power_law_1.2,0.07825279831886292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,1,power_law_1.2,0.10652159452438355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,1,power_law_1.2,0.1375167965888977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,1,power_law_1.2,0.21746559143066407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,1,power_law_1.2,0.22410240173339843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,1,power_law_1.2,0.2856192111968994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,1,power_law_1.2,0.35394558906555174
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,1,power_law_1.2,0.4175871849060059
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,1,power_law_1.2,0.32325119972229005
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,1,power_law_1.2,0.35136001110076903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,1,power_law_1.2,0.43389439582824707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,1,power_law_1.2,0.3384000062942505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,1,power_law_1.2,0.3841344118118286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,1,power_law_1.2,0.4576064109802246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,1,power_law_1.2,0.39343359470367434
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,1,power_law_1.2,0.4439551830291748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,1,power_law_1.2,6.258726501464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,1,power_law_1.2,0.5229695796966553
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,1,power_law_1.2,0.46695680618286134
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,1,power_law_1.2,0.5700032234191894
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,1,power_law_1.01,4.845568084716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,1,power_law_1.01,34.02153015136719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,1,power_law_1.2,0.7816959857940674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,1,power_law_1.2,0.9753408432006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,1,power_law_1.2,28.362060546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,1,power_law_1.01,6.851494598388672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,1,balanced,0.17294933398564658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,1,balanced,0.3055306673049927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,1,balanced,0.5255039930343628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,1,balanced,0.9833333492279053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,1,power_law_1.2,9.226163482666015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,1,balanced,1.8436106046040852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,1,balanced,3.606069246927897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,1,balanced,3.643696149190267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,1,power_law_1.2,1.3606783866882324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,1,balanced,3.6604267756144204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,1,power_law_1.2,21.182847595214845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,1,balanced,3.721861203511556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,1,balanced,3.780122756958008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,1,balanced,3.6975253423055015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,1,power_law_1.01,24.377766418457032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,1,balanced,3.766218821207682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,1,balanced,0.11380267143249512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,1,balanced,0.19273066520690918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,1,balanced,3.7828426361083984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,1,balanced,0.32231465975443524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,1,balanced,0.6061386664708456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,1,balanced,1.1745279630025227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,1,balanced,3.846869468688965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,1,balanced,2.293994744618734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,1,balanced,2.2976907094319663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,1,balanced,4.044143994649251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,1,balanced,2.3302292823791504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,1,balanced,4.071653366088867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,1,balanced,2.389285405476888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,1,balanced,2.360186735788981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,1,balanced,4.127082824707031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,1,balanced,2.3818079630533853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,1,balanced,2.363738695780436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,1,balanced,4.223381360371907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,1,balanced,2.413893381754557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,1,balanced,2.4548586209615073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,1,balanced,4.343690554300944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,1,balanced,2.5379039446512857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,1,balanced,4.6270186106363935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,1,balanced,2.581973393758138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,1,power_law_1.2,1.7581632614135743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,1,power_law_1.01,50.319967651367186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,1,balanced,2.6587732632954917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,1,balanced,4.965338706970215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,1,balanced,2.8688694636027017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,1,balanced,3.0310398737589517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,1,power_law_1.2,12.278514862060547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,1,power_law_1.2,47.2329345703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,1,balanced,7.65779177347819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,1,balanced,3.3194452921549478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,1,balanced,3.645711898803711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,1,balanced,7.800010681152344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,1,balanced,5.726245244344075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,1,balanced,6.5054505666097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,1,balanced,11.667840321858725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,1,balanced,9.341514587402344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,1,balanced,15.854949951171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,1,balanced,12.145909627278646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,1,balanced,17.921541849772137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,1,balanced,25.726938883463543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,1,power_law_1.2,2.5578239440917967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,1,balanced,23.67547607421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,1,balanced,37.005409240722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,1,balanced,0.10413866241772969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,1,balanced,0.1785866618156433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,1,balanced,0.32025599479675293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,1,balanced,0.5715413490931193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,1,balanced,1.1106666723887126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,1,power_law_1.01,0.14608639478683472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,1,power_law_1.01,0.2904191970825195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,1,balanced,2.163877328236898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,1,power_law_1.01,64.19852294921876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,1,power_law_1.01,0.5094719886779785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,1,balanced,2.182352066040039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,1,power_law_1.01,0.9373824119567871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,1,balanced,2.1865386962890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,1,power_law_1.01,1.40960636138916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,1,power_law_1.01,0.11068160533905029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,1,balanced,2.1995253562927246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,1,power_law_1.01,0.1908031940460205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,1,power_law_1.01,1.80383358001709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,1,power_law_1.01,0.3202176094055176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,1,power_law_1.01,0.6037504196166992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,1,power_law_1.01,2.545561599731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,1,power_law_1.01,0.8747072219848633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,1,power_law_1.01,1.1874752044677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,1,balanced,2.196005344390869
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,1,balanced,0.058490668733914696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,1,power_law_1.01,2.7096704483032226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,1,balanced,0.08690667152404785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,1,balanced,0.14120533068974814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,1,balanced,0.2482560078303019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,1,power_law_1.01,1.681056022644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,1,balanced,0.4636213382085164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,1,balanced,0.891050656636556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,1,power_law_1.01,2.8492992401123045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,1,power_law_1.01,1.7687679290771485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,1,power_law_1.01,1.808358383178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,1,power_law_1.01,2.9016191482543947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,1,balanced,2.211146672566732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,1,power_law_1.01,1.8679168701171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,1,balanced,2.222208023071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,1,balanced,0.901413361231486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,1,power_law_1.01,3.125984001159668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,1,balanced,0.9058187007904053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,1,power_law_1.01,1.9474624633789062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,1,balanced,2.2381653785705566
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,1,balanced,0.9083893299102783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,1,balanced,0.9122506777445475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,1,balanced,2.262458642323812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,1,power_law_1.01,2.143974494934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,1,power_law_1.01,3.202617645263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,1,balanced,2.2872427304585776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,1,power_law_1.01,2.2321088790893553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,1,balanced,2.3172426223754883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,1,power_law_1.01,3.3778305053710938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,1,balanced,2.3598292668660483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,1,balanced,0.9204906622568766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,1,power_law_1.01,2.5156991958618162
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,1,balanced,0.9286932945251465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,1,balanced,2.4597439765930176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,1,balanced,0.9396106402079264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,1,power_law_1.01,3.7354625701904296
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,1,balanced,0.9540212949117025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,1,balanced,2.521770636240641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,1,balanced,2.6850452423095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,1,power_law_1.01,3.5059200286865235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,1,balanced,3.0350828170776367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,1,power_law_1.01,2.260371208190918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,1,balanced,1.034821351369222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,1,power_law_1.01,3.685318374633789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,1,balanced,1.0498186747233074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,1,power_law_1.01,2.354528045654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,1,balanced,4.397295951843262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,1,power_law_1.01,3.802694320678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,1,power_law_1.01,2.572102355957031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,1,balanced,5.1515200932820635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,1,power_law_1.2,35.37240295410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,1,balanced,1.081391970316569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,1,power_law_1.01,2.917446327209473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,1,power_law_1.01,4.176204681396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,1,balanced,1.0882399876912434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,1,power_law_1.2,18.171673583984376
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,1,balanced,1.1268479824066162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,1,balanced,7.039194742838542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,1,balanced,1.23416002591451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,1,power_law_1.01,3.256966400146484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,1,balanced,1.2923839886983235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,1,power_law_1.01,4.5033409118652346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,1,balanced,1.5935840606689453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,1,balanced,1.6064906120300293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,1,power_law_1.01,3.918841552734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,1,balanced,9.06284268697103
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,1,balanced,3.246357282002767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,1,power_law_1.01,5.413836669921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,1,balanced,2.1257920265197754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,1,power_law_1.01,4.627193450927734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,1,power_law_1.2,63.50999145507812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,1,balanced,3.1157814661661782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,1,power_law_1.01,6.336345672607422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,1,balanced,4.13596789042155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,1,power_law_1.01,6.076012802124024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,1,balanced,13.086474100748697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,1,power_law_1.01,7.988166046142578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,1,power_law_1.01,7.4920188903808596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,1,balanced,17.134559631347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,1,power_law_1.01,9.678943634033203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,1,power_law_1.2,0.14551039934158325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,1,power_law_1.2,0.2880768060684204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,1,power_law_1.2,0.5053311824798584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,1,power_law_1.2,0.9176128387451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,1,power_law_1.2,3.397689437866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,1,power_law_1.01,10.323820495605469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,1,power_law_1.2,1.2441472053527831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,1,power_law_1.2,1.6909696578979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,1,power_law_1.01,0.056409597396850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,1,power_law_1.01,13.20908203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,1,power_law_1.01,0.08381440043449402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,1,power_law_1.2,2.5285247802734374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,1,power_law_1.01,0.13651200532913207
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,1,power_law_1.01,0.24508159160614013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,1,power_law_1.2,2.7353727340698244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,1,power_law_1.2,2.7976831436157226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,1,power_law_1.01,13.272236633300782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,1,power_law_1.01,0.3586112022399902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,1,power_law_1.2,2.940140724182129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,1,power_law_1.01,0.489350414276123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,1,power_law_1.01,0.669868803024292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,1,power_law_1.2,3.162726402282715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,1,power_law_1.01,16.68053741455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,1,power_law_1.2,3.3044288635253904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,1,power_law_1.01,0.7150976181030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,1,power_law_1.2,3.4711231231689452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,1,power_law_1.01,18.990106201171876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,1,power_law_1.2,3.7853633880615236
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,1,power_law_1.01,0.7204288005828857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,1,power_law_1.01,0.7413695812225342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,1,power_law_1.2,3.772012710571289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,1,power_law_1.01,0.7710015773773193
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,1,power_law_1.01,0.7997439861297607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,1,power_law_1.01,0.8400639533996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,1,power_law_1.2,3.765107345581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,1,power_law_1.01,0.9018239974975586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,1,power_law_1.01,0.98089599609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,1,power_law_1.01,23.369004821777345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,1,power_law_1.01,1.0494720458984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,1,power_law_1.2,3.9988479614257812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,1,power_law_1.01,1.2425408363342285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,1,power_law_1.01,1.1469887733459472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,1,power_law_1.2,4.444601440429688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,1,power_law_1.01,1.3217599868774415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,1,power_law_1.01,1.3224639892578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,1,power_law_1.01,0.10186879634857178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,1,power_law_1.2,4.761523056030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,1,power_law_1.01,1.5437503814697267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,1,power_law_1.01,24.758009338378905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,1,power_law_1.01,0.17555199861526488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,1,power_law_1.01,0.319648003578186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,1,power_law_1.01,1.5833855628967286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,1,power_law_1.01,0.5687168121337891
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,1,power_law_1.01,1.8551103591918945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,1,power_law_1.01,0.8698880195617675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,1,power_law_1.2,5.585734558105469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,1,power_law_1.01,1.141766357421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,1,power_law_1.01,2.131622314453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,1,power_law_1.2,0.10933760404586793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,1,power_law_1.01,1.608121681213379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,1,power_law_1.2,0.19111039638519287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,1,power_law_1.01,2.5710399627685545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,1,power_law_1.2,0.32065279483795167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,1,power_law_1.01,1.6864191055297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,1,power_law_1.2,6.581011199951172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,1,power_law_1.2,0.6034815788269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,1,power_law_1.01,1.742291259765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,1,power_law_1.01,3.4360958099365235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,1,power_law_1.01,1.7702848434448242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,1,power_law_1.01,29.77877197265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,1,power_law_1.01,1.8329280853271483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,1,power_law_1.2,0.8121855735778809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,1,power_law_1.2,7.974297332763672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,1,power_law_1.01,5.2377471923828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,1,power_law_1.01,1.9068992614746094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,1,power_law_1.2,1.0594304084777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,1,power_law_1.01,1.9337919235229493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,1,power_law_1.2,1.6509824752807618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,1,power_law_1.01,1.9750591278076173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,1,power_law_1.2,1.7445119857788085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,1,power_law_1.01,2.0374399185180665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,1,power_law_1.2,1.8069375991821288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,1,power_law_1.2,9.924114990234376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,1,power_law_1.01,2.070444869995117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,1,power_law_1.2,0.10238720178604126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,1,power_law_1.2,0.17512960433959962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,1,power_law_1.01,2.20949764251709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,1,power_law_1.2,0.3186624050140381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,1,power_law_1.2,0.5341695785522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,1,power_law_1.01,2.4357887268066407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,1,power_law_1.2,1.8837503433227538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,1,power_law_1.2,0.7706560134887696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,1,power_law_1.2,2.066099166870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,1,power_law_1.01,2.656537628173828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,1,power_law_1.2,13.348678588867188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,1,power_law_1.2,2.124006462097168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,1,power_law_1.01,3.0952320098876953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,1,power_law_1.2,2.2869247436523437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,1,power_law_1.2,1.072108840942383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,1,power_law_1.01,3.5759681701660155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,1,power_law_1.2,0.05719040036201477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,1,power_law_1.2,2.5604095458984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,1,power_law_1.2,0.0840511977672577
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,1,power_law_1.2,0.1368831992149353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,1,power_law_1.2,2.352799987792969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,1,power_law_1.01,4.588307189941406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,1,power_law_1.2,2.439347267150879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,1,power_law_1.2,1.541062355041504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,1,power_law_1.2,1.633932876586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,1,power_law_1.2,2.654457664489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,1,power_law_1.2,0.21891839504241944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,1,power_law_1.2,16.9198974609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,1,power_law_1.01,5.586457443237305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,1,power_law_1.2,0.320415997505188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,1,power_law_1.2,1.7240447998046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,1,power_law_1.2,0.44911999702453614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,1,power_law_1.2,3.0465024948120116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,1,power_law_1.2,0.6524032115936279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,1,power_law_1.2,1.7668031692504882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,1,power_law_1.2,0.6930496215820312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,1,power_law_1.2,1.8530496597290038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,1,power_law_1.2,3.353094482421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,1,power_law_1.2,0.7122303962707519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,1,power_law_1.01,7.583232116699219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,1,power_law_1.2,1.8707904815673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,1,power_law_1.2,4.055219268798828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,1,power_law_1.2,1.9433536529541016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,1,power_law_1.2,1.9979904174804688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,1,power_law_1.2,0.735206413269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,1,power_law_1.2,0.7581056118011474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,1,power_law_1.2,4.763340759277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,1,power_law_1.2,0.8094464302062988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,1,power_law_1.01,9.690950775146485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,1,power_law_1.2,0.8622847557067871
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,1,power_law_1.2,0.9239104270935059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,1,power_law_1.2,24.377536010742187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,1,power_law_1.2,1.0285056114196778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,1,power_law_1.2,2.0958911895751955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,1,power_law_1.2,6.176729583740235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,1,power_law_1.2,23.56192626953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,1,power_law_1.2,1.1094528198242188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,1,balanced,0.12270399928092957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,1,balanced,0.25917333364486694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,1,balanced,0.4434453248977661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,1,power_law_1.2,2.1794431686401365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,1,power_law_1.2,1.3086848258972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,1,power_law_1.2,1.2209343910217285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,1,power_law_1.2,2.2471872329711915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,1,power_law_1.2,1.4148736000061035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,1,power_law_1.2,7.623622131347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,1,power_law_1.2,2.514316749572754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,1,power_law_1.01,13.713357543945312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,1,balanced,0.806272029876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,1,power_law_1.2,1.395571231842041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,1,balanced,1.5668427149454753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,1,power_law_1.2,1.6416000366210937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,1,power_law_1.2,2.701375961303711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,1,balanced,3.0708373387654624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,1,power_law_1.2,1.6395967483520508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,1,power_law_1.2,3.2447681427001953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,1,balanced,3.1175626118977866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,1,power_law_1.2,1.9246015548706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,1,power_law_1.2,10.505868530273437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,1,balanced,3.1204586029052734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,1,power_law_1.2,2.2542144775390627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,1,balanced,3.118133227030436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,1,balanced,3.131984074910482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,1,power_law_1.2,2.700057601928711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,1,power_law_1.2,3.7262527465820314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,1,power_law_1.01,17.709228515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,1,power_law_1.2,3.599046325683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,1,balanced,0.10220799843470256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,1,power_law_1.2,31.03189697265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,1,balanced,0.1697439948717753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,1,balanced,0.28008000055948895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,1,balanced,0.525328000386556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,1,power_law_1.2,13.413478088378906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,1,balanced,1.0153600374857585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,1,balanced,1.97979736328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,1,power_law_1.2,5.670572662353516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,1,power_law_1.2,4.69156494140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,1,balanced,3.1526613235473633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,1,balanced,3.1249173482259116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,1,balanced,1.999834696451823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,1,power_law_1.2,5.701350402832031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,1,balanced,0.09458133578300476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,1,balanced,0.1527253290017446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,1,balanced,0.28056534131368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,1,balanced,0.5013920068740845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,1,balanced,0.9527413050333658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,1,balanced,1.8583200772603352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,1,power_law_1.2,52.4032958984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,1,balanced,1.9857652982076008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,1,power_law_1.2,19.190675354003908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,1,balanced,1.9903465906778972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,1,power_law_1.2,7.710726165771485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,1,power_law_1.01,0.11951359510421752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,1,balanced,1.9937705993652344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,1,power_law_1.01,0.24916479587554932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,1,balanced,1.9999252955118816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,1,power_law_1.01,0.44193282127380373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,1,balanced,3.15334415435791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,1,balanced,2.071557362874349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,1,power_law_1.01,0.8029248237609863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,1,balanced,2.047658602396647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,1,balanced,3.1513439814249673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,1,power_law_1.01,1.1913984298706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,1,balanced,0.05418666700522105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,1,balanced,0.07734400033950806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,1,balanced,2.034874598185221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,1,balanced,3.4335947036743164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,1,power_law_1.01,1.5661952018737793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,1,power_law_1.2,9.698067474365235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,1,balanced,2.199087937672933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,1,balanced,2.2055306434631348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,1,balanced,3.46503480275472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,1,power_law_1.01,2.2160768508911133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,1,power_law_1.2,5.108505630493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,1,balanced,2.2602079709370932
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,1,balanced,0.12359999616940816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,1,balanced,0.21685866514841715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,1,balanced,3.5095040003458657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,1,balanced,0.40037333965301514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,1,power_law_1.01,2.3434751510620115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,1,balanced,2.4434827168782554
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,1,balanced,0.7675039768218994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,1,balanced,1.866378625233968
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,1,balanced,0.774336020151774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,1,balanced,0.7804853121439616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,1,balanced,1.8714720408121746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,1,balanced,3.5761706034342446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,1,balanced,2.5463573137919107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,1,power_law_1.01,2.416761589050293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,1,balanced,0.7822079658508301
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,1,balanced,0.78603196144104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,1,balanced,1.8805759747823079
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,1,balanced,0.7931626637776693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,1,balanced,2.8049494425455728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,1,balanced,3.704106648763021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,1,balanced,1.8842719395955403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,1,power_law_1.01,2.4944896697998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,1,power_law_1.2,24.973178100585937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,1,balanced,1.9005813598632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,1,balanced,3.1088587443033853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,1,power_law_1.2,13.725932312011718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,1,balanced,3.9173386891682944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,1,balanced,1.9248107274373372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,1,power_law_1.01,2.5750207901000977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,1,power_law_1.01,0.09896960258483886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,1,power_law_1.01,0.17096320390701295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,1,balanced,1.9319626490275066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,1,power_law_1.01,0.2793215990066528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,1,balanced,4.913573265075684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,1,balanced,0.801695982615153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,1,balanced,1.9455787340799968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,1,power_law_1.01,2.728563117980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,1,balanced,0.8081706364949545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,1,power_law_1.01,0.5224959850311279
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,1,balanced,0.822767972946167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,1,balanced,1.9677759806315105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,1,power_law_1.01,0.7939583778381347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,1,balanced,1.9909920692443848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,1,power_law_1.01,2.9542591094970705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,1,balanced,5.545701344807942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,1,power_law_1.01,1.0309951782226563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,1,balanced,2.036874612172445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,1,balanced,4.249530792236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,1,balanced,2.1165547370910645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,1,balanced,0.890181303024292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,1,balanced,2.193648020426432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,1,balanced,0.9024213155110677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,1,balanced,8.061610539754232
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,1,balanced,0.9284266630808512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,1,balanced,6.001509348551433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,1,balanced,0.9337546825408936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,1,balanced,0.9686506589253744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,1,power_law_1.01,1.4833791732788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,1,power_law_1.2,17.72213134765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,1,power_law_1.01,3.1566144943237306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,1,balanced,1.0658506552378337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,1,power_law_1.01,1.4940095901489259
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,1,balanced,1.1124693552652996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,1,balanced,6.773333231608073
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,1,balanced,1.4867839813232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,1,power_law_1.01,1.5934080123901366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,1,balanced,2.3311413129170737
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,1,balanced,1.3606665929158528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,1,balanced,10.494010925292969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,1,power_law_1.01,1.5904704093933106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,1,balanced,2.552245299021403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,1,balanced,2.9193760553995767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,1,power_law_1.01,1.7103551864624023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,1,balanced,1.8484959602355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,1,balanced,3.8161067962646484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,1,power_law_1.01,3.0427839279174806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,1,balanced,9.843584060668945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,1,balanced,2.646442731221517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,1,balanced,4.1637013753255205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,1,power_law_1.01,3.1671424865722657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,1,balanced,3.43669859568278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,1,power_law_1.01,1.8602815628051759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,1,balanced,15.472522735595703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,1,power_law_1.01,3.2722942352294924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,1,balanced,6.034746805826823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,1,power_law_1.01,1.9095935821533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,1,power_law_1.01,0.09502080082893372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,1,balanced,13.47421391805013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,1,power_law_1.01,0.15111680030822755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,1,power_law_1.01,2.141299247741699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,1,power_law_1.01,0.2797823905944824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,1,balanced,7.834805170694987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,1,power_law_1.01,0.4954944133758545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,1,power_law_1.01,1.9303295135498046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,1,power_law_1.01,0.7457727909088134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,1,power_law_1.01,2.030463981628418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,1,power_law_1.01,3.547903823852539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,1,power_law_1.01,1.0015935897827148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,1,power_law_1.01,2.1783552169799805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,1,power_law_1.01,1.397753620147705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,1,power_law_1.01,3.736774444580078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,1,power_law_1.01,1.4684096336364747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,1,power_law_1.01,2.5374143600463865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,1,balanced,20.548442840576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,1,balanced,11.307722727457682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,1,power_law_1.01,1.49749755859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,1,power_law_1.2,0.12215039730072022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,1,power_law_1.01,2.8022079467773438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,1,power_law_1.01,1.5376319885253906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,1,power_law_1.2,0.26350719928741456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,1,power_law_1.2,0.4528512001037598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,1,power_law_1.01,1.5876223564147949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,1,power_law_1.2,0.7978559970855713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,1,power_law_1.01,0.052537602186203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,1,power_law_1.01,3.3962238311767576
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,1,power_law_1.01,0.07612159848213196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,1,power_law_1.01,1.639276885986328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,1,power_law_1.2,1.0599871635437013
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,1,power_law_1.01,0.12194559574127198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,1,power_law_1.01,4.6148223876953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,1,power_law_1.01,0.21469440460205078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,1,power_law_1.01,1.6510847091674805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,1,power_law_1.2,1.453343963623047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,1,balanced,22.133082071940105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,1,power_law_1.01,4.0256702423095705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,1,balanced,14.89908218383789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,1,power_law_1.01,1.6633983612060548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,1,power_law_1.2,2.1775360107421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,1,power_law_1.01,1.7389312744140626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,1,power_law_1.2,0.10035200119018554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,1,power_law_1.01,5.199833679199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,1,power_law_1.2,0.16872960329055786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,1,power_law_1.2,2.3020416259765626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,1,power_law_1.01,5.205452728271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,1,power_law_1.01,1.7713855743408202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,1,power_law_1.2,0.2800832033157349
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,1,power_law_1.01,0.3156800031661987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,1,power_law_1.2,0.5196095943450928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,1,power_law_1.01,1.9025407791137696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,1,power_law_1.2,2.3980031967163087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,1,power_law_1.2,0.6974592208862305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,1,power_law_1.01,2.1145984649658205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,1,power_law_1.01,6.838771057128906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,1,power_law_1.2,2.459507179260254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,1,power_law_1.01,6.459161376953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,1,power_law_1.01,0.4153600215911865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,1,power_law_1.2,2.622336006164551
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,1,power_law_1.01,0.5744063854217529
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,1,power_law_1.01,0.6033152103424072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,1,power_law_1.2,0.9244416236877442
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,1,power_law_1.01,0.626527976989746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,1,power_law_1.2,2.795916748046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,1,power_law_1.01,8.223999786376954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,1,power_law_1.2,1.4319680213928223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,1,power_law_1.01,2.311123275756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,1,balanced,28.913599650065105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,1,power_law_1.2,2.870963287353516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,1,power_law_1.2,1.4732928276062012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,1,power_law_1.01,2.667091178894043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,1,power_law_1.2,1.5470720291137696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,1,power_law_1.2,3.2880958557128905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,1,power_law_1.01,0.6252543926239014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,1,power_law_1.2,1.5870016098022461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,1,power_law_1.01,3.1433855056762696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,1,power_law_1.01,0.6686079978942872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,1,power_law_1.01,8.922777557373047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,1,power_law_1.01,0.6876992225646973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,1,power_law_1.2,3.0671360015869142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,1,power_law_1.01,0.7265984058380127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,1,power_law_1.01,11.618956756591796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,1,power_law_1.01,3.9732608795166016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,1,power_law_1.01,0.7760704040527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,1,power_law_1.2,3.2716094970703127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,1,power_law_1.01,0.839628791809082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,1,power_law_1.01,0.8998720169067382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,1,power_law_1.2,1.759244728088379
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,1,power_law_1.01,1.04585599899292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,1,power_law_1.2,3.4000640869140626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,1,power_law_1.01,4.872236633300782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,1,power_law_1.2,0.09405440092086792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,1,power_law_1.2,1.8681663513183593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,1,power_law_1.01,0.9789119720458984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,1,power_law_1.2,0.15262720584869385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,1,power_law_1.01,11.44698257446289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,1,power_law_1.2,0.2799936056137085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,1,power_law_1.01,1.1392512321472168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,1,power_law_1.2,0.47146878242492674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,1,power_law_1.01,1.1209152221679688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,1,power_law_1.2,0.7063295841217041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,1,power_law_1.01,6.585497283935547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,1,power_law_1.01,1.3197759628295898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,1,power_law_1.2,0.9572799682617188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,1,power_law_1.01,14.297517395019531
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,1,power_law_1.01,1.3599103927612304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,1,power_law_1.2,1.9940671920776367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,1,power_law_1.2,1.3378368377685548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,1,power_law_1.2,3.794297790527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,1,power_law_1.01,1.6076671600341796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,1,power_law_1.2,1.4326656341552735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,1,power_law_1.2,1.4925248146057128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,1,power_law_1.2,4.001433563232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,1,power_law_1.01,8.36084442138672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,1,power_law_1.2,1.533676815032959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,1,power_law_1.2,2.2272127151489256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,1,power_law_1.01,16.400051879882813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,1,power_law_1.2,1.578444766998291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,1,power_law_1.2,4.6526336669921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,1,power_law_1.01,1.8285440444946288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,1,power_law_1.2,1.6437376022338868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,1,power_law_1.2,2.054150390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,1,power_law_1.2,1.6708671569824218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,1,power_law_1.2,2.1237503051757813
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,1,power_law_1.01,2.2254144668579103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,1,power_law_1.2,1.715827178955078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,1,power_law_1.2,5.4224193572998045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,1,power_law_1.2,2.308012771606445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,1,power_law_1.01,20.427276611328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,1,power_law_1.01,11.84985580444336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,1,power_law_1.2,1.802297592163086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,1,power_law_1.01,2.9794111251831055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,1,power_law_1.2,2.596646308898926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,1,power_law_1.2,1.8366207122802733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,1,power_law_1.2,1.9532928466796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,1,power_law_1.2,6.7504638671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,1,power_law_1.01,4.768211364746094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,1,power_law_1.2,2.187295913696289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,1,power_law_1.2,2.4111616134643556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,1,power_law_1.2,2.889606475830078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,1,power_law_1.2,8.6914306640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,1,power_law_1.2,2.7934591293334963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,1,power_law_1.01,15.36810302734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,1,power_law_1.2,3.496384048461914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,1,power_law_1.01,21.384632873535157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,1,power_law_1.2,3.215462493896484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,1,balanced,0.029365333418051403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,1,balanced,0.04182933270931244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,1,balanced,0.06734399994214375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,1,balanced,0.10469866792360942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,1,balanced,0.1676959991455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,1,balanced,0.16980266571044922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,1,balanced,0.17041067282358804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,1,balanced,0.17215466499328613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,1,balanced,0.17315733432769775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,1,balanced,0.17348267634709677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,1,balanced,0.17392534017562866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,1,balanced,0.1927573283513387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,1,balanced,0.19364267587661743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,1,power_law_1.2,4.081075286865234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,1,balanced,0.19946666558583578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,1,balanced,0.20534932613372803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,1,power_law_1.01,25.786868286132812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,1,power_law_1.2,11.491725158691406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,1,power_law_1.2,4.118041610717773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,1,balanced,0.033520000676314034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,1,balanced,0.041573333243529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,1,balanced,0.05620799958705902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,1,power_law_1.2,4.976966476440429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,1,balanced,0.21075733502705893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,1,balanced,0.2230986754099528
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,1,power_law_1.2,0.05264000296592712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,1,balanced,0.2513333360354106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,1,balanced,0.27906666199366253
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,1,power_law_1.2,0.07542399764060974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,1,balanced,0.41699198881785077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,1,power_law_1.2,5.338803100585937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,1,balanced,0.475818673769633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,1,power_law_1.2,0.1211135983467102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,1,balanced,0.6694026788075765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,1,power_law_1.2,0.191212797164917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,1,balanced,0.8663199742635092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,1,balanced,0.09088533123334248
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,1,power_law_1.2,0.28677759170532224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,1,balanced,0.13927466670672098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,1,balanced,1.2591626644134521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,1,power_law_1.2,6.717139434814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,1,power_law_1.2,0.3949631929397583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,1,balanced,1.653488000233968
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,1,power_law_1.2,0.5703551769256592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,1,balanced,2.4404853185017905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,1,power_law_1.2,0.6042496204376221
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,1,power_law_1.2,0.613036823272705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,1,balanced,0.027488000690937042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,1,balanced,3.225301424662272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,1,balanced,0.14814933141072592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,1,balanced,0.14224533240000406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,1,balanced,0.14380266269048056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,1,balanced,0.14405866463979086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,1,balanced,0.1453279952208201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,1,balanced,0.1481706698735555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,1,balanced,0.15044800440470377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,1,balanced,0.15359999736150107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,1,balanced,0.1586133340994517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,1,balanced,0.2099519968032837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,1,power_law_1.2,14.49473876953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,1,power_law_1.2,6.55137939453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,1,power_law_1.2,0.6457791805267334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,1,balanced,0.025439999997615814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,1,balanced,0.029663999875386555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,1,balanced,0.03530666728814443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,1,balanced,0.05412800113360087
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,1,balanced,0.07830933233102162
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,1,balanced,0.08038400113582611
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,1,balanced,0.08337066570917766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,1,balanced,0.08480532964070638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,1,balanced,0.08515200018882751
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,1,balanced,0.08675733208656311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,1,balanced,0.21492799123128256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,1,balanced,0.0888266662756602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,1,balanced,0.21778666973114014
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,1,balanced,0.09497066338857015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,1,balanced,0.2400533358256022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,1,balanced,0.25967466831207275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,1,balanced,0.3962453206380208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,1,balanced,0.43854931990305585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,1,power_law_1.2,9.065062713623046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,1,balanced,0.6135253508885702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,1,power_law_1.2,0.6751232147216797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,1,balanced,0.7977440357208252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,1,power_law_1.2,8.48958740234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,1,balanced,1.1577760378519695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,1,power_law_1.2,0.6867712020874024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,1,balanced,1.5245386759440105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,1,balanced,0.09886399904886882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,1,balanced,0.10017066200574239
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,1,balanced,0.0978613297144572
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,1,balanced,0.10213333368301392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,1,balanced,0.10528000195821126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,1,balanced,2.2337652842203775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,1,balanced,0.12076266606648763
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,1,balanced,0.12610133488972983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,1,balanced,0.15958933035532633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,1,balanced,0.15639999508857727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,1,balanced,0.2989013393719991
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,1,balanced,0.24514667193094888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,1,balanced,2.9679892857869468
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,1,balanced,0.3428479830423991
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,1,balanced,0.43572266896565753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,1,power_law_1.2,19.784051513671876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,1,balanced,0.61736532052358
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,1,power_law_1.2,0.7180223941802979
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,1,balanced,0.81223464012146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,1,power_law_1.2,0.8019007682800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,1,power_law_1.2,11.564710235595703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,1,balanced,0.04301866888999939
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,1,power_law_1.2,0.8681407928466797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,1,power_law_1.2,0.9517248153686524
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,1,power_law_1.2,1.1117695808410644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,1,power_law_1.01,0.02584959864616394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,1,power_law_1.01,0.041791999340057374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,1,power_law_1.01,0.06470400094985962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,1,power_law_1.01,0.08440319895744323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,1,power_law_1.01,0.10396159887313842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,1,power_law_1.01,0.14294400215148925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,1,power_law_1.2,11.975603485107422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,1,power_law_1.01,0.15150079727172852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,1,power_law_1.01,0.15442559719085694
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,1,power_law_1.2,1.0412863731384276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,1,power_law_1.2,1.209887981414795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,1,power_law_1.2,16.518707275390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,1,power_law_1.01,0.025894400477409363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,1,power_law_1.01,0.1665536046028137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,1,power_law_1.2,1.1853695869445802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,1,power_law_1.01,0.1740671992301941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,1,balanced,0.0641599992911021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,1,power_law_1.01,0.1904960036277771
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,1,power_law_1.2,1.397273635864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,1,power_law_1.2,26.269573974609376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,1,power_law_1.01,0.03149439990520477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,1,power_law_1.01,0.18715519905090333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,1,power_law_1.01,0.041388800740242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,1,power_law_1.01,0.19109760522842406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,1,power_law_1.01,0.054067200422286986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,1,power_law_1.01,0.20723199844360352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,1,power_law_1.2,1.4145536422729492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,1,power_law_1.01,0.21521921157836915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,1,power_law_1.01,0.07599999904632568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,1,power_law_1.01,0.22720639705657958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,1,power_law_1.01,0.0933184027671814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,1,power_law_1.01,0.2490623950958252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,1,power_law_1.2,1.6499391555786134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,1,power_law_1.01,0.2960319995880127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,1,power_law_1.01,0.3464384078979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,1,power_law_1.01,0.4410111904144287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,1,power_law_1.2,15.453158569335937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,1,power_law_1.2,1.944268798828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,1,power_law_1.01,0.5386559963226318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,1,power_law_1.01,0.7341375827789307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,1,power_law_1.01,0.1265920042991638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,1,power_law_1.2,2.3271039962768554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,1,power_law_1.2,21.534906005859376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,1,balanced,0.11936533451080322
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,1,power_law_1.2,3.047974395751953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,1,power_law_1.01,0.9295807838439941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,1,power_law_1.01,0.04085760116577149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,1,power_law_1.01,0.13115520477294923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,1,power_law_1.01,0.1334015965461731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,1,power_law_1.01,1.3262911796569825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,1,power_law_1.01,0.13640320301055908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,1,power_law_1.01,0.024607999622821806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,1,power_law_1.01,0.14088959693908693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,1,power_law_1.01,0.030041599273681642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,1,power_law_1.01,0.14365439414978026
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,1,power_law_1.01,0.03656319975852966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,1,power_law_1.01,0.1558784008026123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,1,power_law_1.01,0.046777600049972536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,1,power_law_1.01,0.16120959520339967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,1,power_law_1.01,0.05566080212593079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,1,power_law_1.01,0.18057600259780884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,1,power_law_1.01,0.07160320281982421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,1,power_law_1.01,0.19930239915847778
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,1,power_law_1.01,0.07626240253448487
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,1,power_law_1.01,0.07941120266914367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,1,power_law_1.01,0.08112639784812928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,1,power_law_1.01,0.08305280208587647
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,1,power_law_1.01,0.08664320111274719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,1,power_law_1.01,1.7197824478149415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,1,power_law_1.2,7.328511810302734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,1,power_law_1.01,0.09070079922676086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,1,power_law_1.01,0.09726079702377319
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,1,power_law_1.01,0.10723199844360351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,1,power_law_1.2,0.034585601091384886
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,1,power_law_1.01,0.09527680277824402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,1,power_law_1.01,2.505088043212891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,1,power_law_1.2,0.05614719986915588
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,1,power_law_1.01,0.10077439546585083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,1,power_law_1.01,0.2214400053024292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,1,power_law_1.2,0.06632959842681885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,1,power_law_1.01,0.11329280138015747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,1,power_law_1.01,0.2342400074005127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,1,power_law_1.2,0.08266239762306213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,1,power_law_1.01,0.2742719888687134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,1,power_law_1.2,0.09875199794769288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,1,power_law_1.2,0.15063680410385133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,1,power_law_1.01,0.316979193687439
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,1,power_law_1.2,4.959072113037109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,1,power_law_1.01,0.4038271903991699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,1,power_law_1.01,0.062054401636123656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,1,power_law_1.01,0.49482240676879885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,1,power_law_1.01,0.6702208042144775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,1,power_law_1.2,0.030188798904418945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,1,balanced,0.2023306687672933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,1,power_law_1.2,0.02568959891796112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,1,balanced,0.2067520022392273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,1,power_law_1.01,0.8469696044921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,1,power_law_1.01,3.290163040161133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,1,power_law_1.2,0.15731199979782104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,1,power_law_1.2,0.1624511957168579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,1,power_law_1.2,0.16992640495300293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,1,power_law_1.2,0.17808640003204346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,1,power_law_1.2,0.19417599439620972
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,1,power_law_1.01,0.11921919584274292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,1,power_law_1.2,0.04121600091457367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,1,power_law_1.2,0.19207680225372314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,1,power_law_1.01,0.13716479539871215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,1,power_law_1.2,0.19831039905548095
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,1,power_law_1.01,0.14949120283126832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,1,power_law_1.2,0.21057920455932616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,1,power_law_1.01,1.2009599685668946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,1,power_law_1.01,0.174399995803833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,1,power_law_1.2,0.21899518966674805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,1,power_law_1.2,0.02341119945049286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,1,power_law_1.01,0.22926719188690187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,1,power_law_1.2,0.23220479488372803
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,1,power_law_1.2,0.028595200181007384
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,1,power_law_1.01,0.2791680097579956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,1,power_law_1.01,1.558572769165039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,1,power_law_1.2,0.03530240058898926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,1,power_law_1.01,0.3727423906326294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,1,power_law_1.2,0.04347519874572754
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,1,power_law_1.2,0.05394560098648071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,1,power_law_1.2,0.05064319968223572
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,1,power_law_1.2,0.07247359752655029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,1,power_law_1.01,2.2644800186157226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,1,power_law_1.2,0.07645440101623535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,1,power_law_1.2,0.0692031979560852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,1,power_law_1.2,0.07863039970397949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,1,power_law_1.2,0.08831999897956848
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,1,power_law_1.2,0.08199040293693542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,1,power_law_1.2,0.1297919988632202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,1,power_law_1.2,0.08290560245513916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,1,power_law_1.2,0.2567487955093384
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,1,power_law_1.2,0.08773760199546814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,1,power_law_1.01,2.9698112487792967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,1,power_law_1.2,0.091839998960495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,1,balanced,0.07981333136558533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,1,balanced,0.13226133584976196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,1,power_law_1.2,0.09752320051193238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,1,power_law_1.01,0.4674431800842285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,1,power_law_1.2,0.10913920402526855
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,1,power_law_1.2,0.09778559803962708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,1,power_law_1.01,0.6483071804046631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,1,power_law_1.2,0.040652799606323245
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,1,power_law_1.2,0.10402560234069824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,1,power_law_1.2,0.11592320203781128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,1,power_law_1.01,0.8642560005187988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,1,power_law_1.2,0.12547199726104735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,1,balanced,0.06431999802589417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,1,power_law_1.2,0.13058559894561766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,1,balanced,0.10658133029937744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,1,balanced,0.16592533389727274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,1,power_law_1.2,0.1351807951927185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,1,power_law_1.2,0.30369279384613035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,1,power_law_1.2,0.1388416051864624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,1,balanced,0.23895466327667236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,1,power_law_1.2,0.35304319858551025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,1,balanced,0.44268798828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,1,power_law_1.2,0.44714879989624023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,1,power_law_1.01,0.09475839734077454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,1,power_law_1.2,0.5458752155303955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,1,power_law_1.2,0.14499839544296264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,1,balanced,0.20747200647989908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,1,power_law_1.2,0.15144959688186646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,1,balanced,0.2079520026842753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,1,balanced,0.29019733270009357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,1,balanced,0.21038933595021567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,1,balanced,0.5346560080846151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,1,balanced,0.5354826847712199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,1,power_law_1.2,0.1409600019454956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,1,balanced,0.5372906525929769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,1,balanced,0.790229320526123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,1,balanced,0.5404373407363892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,1,power_law_1.2,0.17754240036010743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,1,balanced,0.7912639776865641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,1,balanced,0.5414666732152303
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,1,power_law_1.2,0.23644158840179444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,1,balanced,0.5425653457641602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,1,balanced,0.7958292961120605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,1,balanced,0.5469333330790201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,1,balanced,0.7954293092091879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,1,balanced,0.6181120077768961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,1,balanced,0.8012479941050211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,1,balanced,0.6304213205973307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,1,balanced,0.8015039761861166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,1,balanced,0.6472746531168619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,1,balanced,0.8117492993672689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,1,balanced,0.6691093444824219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,1,balanced,0.8986453215281168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,1,balanced,0.6883947054545084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,1,power_law_1.2,0.14897279739379882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,1,balanced,0.7172746658325195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,1,power_law_1.2,0.05974400043487549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,1,power_law_1.2,0.1580672025680542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,1,balanced,0.7913973331451416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,1,power_law_1.2,0.16254080533981324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,1,balanced,0.8744746843973795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,1,power_law_1.2,0.18264960050582885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,1,power_law_1.2,0.20227200984954835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,1,power_law_1.2,0.22392959594726564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,1,power_law_1.2,0.2842303991317749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,1,power_law_1.2,0.2351680040359497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,1,power_law_1.2,0.278438401222229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,1,power_law_1.2,0.7439295768737793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,1,balanced,0.9046560128529867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,1,power_law_1.2,0.3240000009536743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,1,balanced,0.9135200182596842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,1,power_law_1.2,0.4137599945068359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,1,balanced,0.9274186293284098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,1,power_law_1.2,0.49733757972717285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,1,balanced,0.9377333323160807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,1,balanced,1.3715786933898926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,1,balanced,0.9621013005574545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,1,balanced,1.0128693580627441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,1,balanced,1.5520426432291667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,1,power_law_1.01,0.1189695954322815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,1,power_law_1.2,0.6790207862854004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,1,power_law_1.01,0.17664639949798583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,1,balanced,2.2163413365681968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,1,balanced,0.21091733376185098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,1,balanced,2.898325284322103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,1,power_law_1.2,0.9360256195068359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,1,balanced,1.0770986874898274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,1,balanced,4.245205243428548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,1,balanced,1.5133867263793945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,1,balanced,1.6685546239217122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,1,balanced,2.3860586484273276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,1,balanced,5.602912267049153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,1,power_law_1.2,0.8533760070800781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,1,balanced,3.4539305369059243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,1,power_law_1.2,1.3305536270141602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,1,balanced,4.608250617980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,1,power_law_1.2,0.09052799940109253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,1,balanced,8.284634908040365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,1,balanced,6.038501103719075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,1,power_law_1.01,0.1825600028038025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,1,balanced,0.2118613322575887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,1,power_law_1.01,0.1878399968147278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,1,power_law_1.2,1.2067839622497558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,1,power_law_1.01,0.19174400568008423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,1,power_law_1.2,0.37879679203033445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,1,balanced,10.984959920247396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,1,balanced,9.683418909708658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,1,balanced,0.21397866805394491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,1,balanced,0.062074666221936546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,1,balanced,0.21735467513402304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,1,balanced,0.10169600447018941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,1,balanced,0.1605226695537567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,1,balanced,0.2874880035718282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,1,balanced,0.5194559892018636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,1,balanced,0.5225760142008463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,1,balanced,0.5273653268814087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,1,balanced,0.5321706533432007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,1,power_law_1.2,1.728223991394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,1,balanced,13.895263671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,1,balanced,0.5358080069224039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,1,power_law_1.2,0.10803840160369874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,1,balanced,0.5368800163269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,1,power_law_1.2,0.1772096037864685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,1,balanced,0.5454506476720175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,1,balanced,0.5522400140762329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,1,balanced,0.5597920020421346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,1,balanced,0.573312004407247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,1,balanced,0.5836960077285767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,1,balanced,0.5974026521046957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,1,balanced,0.6242186625798544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,1,power_law_1.01,0.1986240029335022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,1,balanced,0.6700747013092041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,1,balanced,0.725770632425944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,1,balanced,1.114181359608968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,1,balanced,1.224992036819458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,1,power_law_1.2,0.4743743896484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,1,balanced,1.752618630727132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,1,balanced,0.22168533007303873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,1,balanced,0.27694400151570636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,1,balanced,2.2814879417419434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,1,balanced,0.2814133365948995
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,1,balanced,0.03379199902216593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,1,balanced,0.054133335749308266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,1,balanced,0.07761066655317943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,1,balanced,0.12474133570988973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,1,balanced,0.21813867489496866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,1,balanced,0.22183465957641602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,1,balanced,3.33351993560791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,1,balanced,0.2253920038541158
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,1,balanced,0.2270453373591105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,1,balanced,0.2305226723353068
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,1,balanced,0.23216533660888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,1,balanced,0.23679999510447183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,1,balanced,0.25763734181722003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,1,balanced,0.26289600133895874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,1,balanced,4.371839841206868
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,1,balanced,0.26971733570098877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,1,balanced,0.2677119970321655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,1,balanced,0.27131734291712445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,1,balanced,0.2834133307139079
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,1,balanced,0.3041119972864787
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,1,balanced,0.32016533613204956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,1,balanced,0.43965331713358563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,1,balanced,0.3788693348566691
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,1,balanced,0.9600319862365723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,1,balanced,6.4429066975911455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,1,balanced,0.5147306521733602
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,1,balanced,0.7334826787312826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,1,balanced,0.2533866763114929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,1,balanced,0.2677066723505656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,1,power_law_1.2,1.5620991706848144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,1,balanced,0.2832159996032715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,1,power_law_1.01,0.2102207899093628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,1,power_law_1.01,0.21644799709320067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,1,power_law_1.01,0.22005760669708252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,1,power_law_1.2,0.18350720405578613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,1,power_law_1.2,0.18935680389404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,1,balanced,8.498565038045248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,1,power_law_1.2,0.197817599773407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,1,balanced,0.9424586296081543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,1,power_law_1.01,0.07445120215415954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,1,power_law_1.01,0.1265920042991638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,1,balanced,1.3610933621724446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,1,power_law_1.01,0.23133440017700196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,1,power_law_1.01,0.3321471929550171
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,1,balanced,1.7914187113444011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,1,power_law_1.01,0.40931200981140137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,1,power_law_1.01,0.0625216007232666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,1,power_law_1.01,0.6723072052001953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,1,power_law_1.01,0.10551040172576905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,1,power_law_1.01,0.16330879926681519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,1,power_law_1.01,0.7035903930664062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,1,power_law_1.01,0.23328640460968017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,1,power_law_1.01,0.7265344142913819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,1,power_law_1.01,0.29719679355621337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,1,power_law_1.01,0.7689792156219483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,1,power_law_1.01,0.4532032012939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,1,power_law_1.01,0.4865856170654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,1,power_law_1.01,0.7786687850952149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,1,power_law_1.01,0.49636478424072267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,1,power_law_1.01,0.8128191947937011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,1,power_law_1.01,0.5366144180297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,1,power_law_1.01,0.8730879783630371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,1,power_law_1.01,0.551091194152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,1,power_law_1.01,0.8809727668762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,1,balanced,0.34359999497731525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,1,balanced,0.37275731563568115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,1,power_law_1.01,0.9036800384521484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,1,power_law_1.01,0.9393343925476074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,1,power_law_1.01,0.23943679332733153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,1,power_law_1.01,0.9464320182800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,1,power_law_1.01,0.2594559907913208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,1,power_law_1.01,1.0019200325012207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,1,power_law_1.01,0.6059775829315186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,1,power_law_1.2,0.20012800693511962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,1,power_law_1.01,0.609984016418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,1,power_law_1.01,1.1345791816711426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,1,power_law_1.01,0.6242752075195312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,1,power_law_1.01,1.3014464378356934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,1,power_law_1.01,0.6575551986694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,1,power_law_1.01,0.7088704109191895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,1,power_law_1.01,1.6052799224853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,1,power_law_1.01,0.2893759965896606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,1,power_law_1.01,0.7390272140502929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,1,power_law_1.01,0.8234560012817382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,1,power_law_1.01,1.9311424255371095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,1,power_law_1.01,0.9792639732360839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,1,power_law_1.01,1.134175968170166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,1,power_law_1.01,2.7939327239990233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,1,power_law_1.01,1.4689344406127929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,1,power_law_1.01,3.352377700805664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,1,power_law_1.01,1.7826431274414063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,1,power_law_1.01,2.4713279724121096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,1,power_law_1.01,0.2629695892333984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,1,power_law_1.01,4.898982238769531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,1,power_law_1.01,0.28594560623168946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,1,power_law_1.2,2.5269760131835937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,1,power_law_1.01,3.1279232025146486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,1,power_law_1.2,0.2103935956954956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,1,power_law_1.2,0.2171072006225586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,1,power_law_1.2,0.22387199401855468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,1,power_law_1.01,4.468896102905274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,1,power_law_1.01,6.452864074707032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,1,power_law_1.01,0.3145535945892334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,1,power_law_1.01,5.817171096801758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,1,power_law_1.01,9.62771224975586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,1,power_law_1.01,8.493587493896484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,1,power_law_1.2,0.2462912082672119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,1,power_law_1.2,0.2643136024475098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,1,power_law_1.01,12.77447052001953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,1,power_law_1.2,0.293503999710083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,1,power_law_1.01,0.37582719326019287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,1,power_law_1.01,0.4258111953735352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,1,power_law_1.01,11.190828704833985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,1,power_law_1.01,0.061427199840545656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,1,power_law_1.01,0.10029439926147461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,1,power_law_1.01,0.15756800174713134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,1,power_law_1.01,0.2251136064529419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,1,power_law_1.01,0.30300800800323485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,1,balanced,0.47861866156260174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,1,power_law_1.01,0.4434944152832031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,1,power_law_1.01,0.46675839424133303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,1,power_law_1.01,0.4838719844818115
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,1,power_law_1.2,0.6607744216918945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,1,power_law_1.01,0.4953023910522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,1,power_law_1.01,0.5021759986877441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,1,power_law_1.01,0.5196352005004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,1,power_law_1.2,0.2645375967025757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,1,balanced,0.5308266480763754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,1,power_law_1.2,0.2940160036087036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,1,balanced,0.8249759674072266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,1,power_law_1.2,0.32635519504547117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,1,balanced,0.9351466496785482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,1,power_law_1.01,0.03292160034179688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,1,balanced,1.336154619852702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,1,power_law_1.01,0.05307520031929016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,1,balanced,1.7329492568969727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,1,power_law_1.01,0.07612159848213196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,1,power_law_1.01,0.10526080131530761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,1,power_law_1.01,0.5230207920074463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,1,power_law_1.01,0.12919039726257325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,1,power_law_1.01,0.5408448219299317
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,1,power_law_1.01,0.19152640104293822
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,1,power_law_1.01,0.20186879634857177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,1,power_law_1.01,0.5735231876373291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,1,power_law_1.01,0.20892159938812255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,1,power_law_1.2,2.2714624404907227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,1,power_law_1.01,0.21731200218200683
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,1,power_law_1.01,0.21809279918670654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,1,power_law_1.01,0.23406078815460205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,1,power_law_1.2,0.07481600046157837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,1,power_law_1.01,0.24666879177093506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,1,power_law_1.2,0.12684799432754518
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,1,power_law_1.01,0.26263039112091063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,1,power_law_1.2,0.22001280784606933
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,1,power_law_1.01,0.2950272083282471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,1,power_law_1.2,0.3052031993865967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,1,power_law_1.01,0.261081600189209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,1,power_law_1.2,0.3810816049575806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,1,power_law_1.01,0.5983808040618896
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,1,power_law_1.01,0.28167679309844973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,1,power_law_1.2,0.6759168148040772
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,1,power_law_1.01,0.31791999340057375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,1,power_law_1.01,0.6214144229888916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,1,power_law_1.2,0.7072896003723145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,1,power_law_1.01,0.31560959815979006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,1,power_law_1.01,0.6744063854217529
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,1,power_law_1.01,0.37404160499572753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,1,power_law_1.2,0.7415552139282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,1,power_law_1.01,0.7955776214599609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,1,power_law_1.2,0.7726784229278565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,1,power_law_1.01,0.9189311981201171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,1,power_law_1.2,0.8003328323364258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,1,power_law_1.2,0.38338561058044435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,1,power_law_1.01,1.174783992767334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,1,power_law_1.01,1.4385727882385253
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,1,power_law_1.01,0.3795520067214966
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,1,power_law_1.01,0.439635181427002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,1,power_law_1.01,1.9566080093383789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,1,power_law_1.01,0.5204415798187256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,1,power_law_1.2,0.8313152313232421
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,1,power_law_1.01,0.6230527877807617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,1,power_law_1.01,2.4735551834106446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,1,power_law_1.01,0.8203647613525391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,1,power_law_1.2,0.437337589263916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,1,power_law_1.01,1.040384006500244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,1,power_law_1.01,3.497420883178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,1,power_law_1.01,1.4553600311279298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,1,power_law_1.2,0.8712575912475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,1,power_law_1.01,2.2006784439086915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,1,power_law_1.2,0.8882431983947754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,1,power_law_1.2,0.06236799955368042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,1,power_law_1.01,4.524953460693359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,1,power_law_1.2,0.9185152053833008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,1,power_law_1.2,0.10491520166397095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,1,power_law_1.2,0.16110080480575562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,1,power_law_1.2,0.9352640151977539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,1,power_law_1.2,0.20183680057525635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,1,power_law_1.2,0.9733375549316406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,1,power_law_1.2,0.26590719223022463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,1,power_law_1.2,0.44263038635253904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,1,power_law_1.2,1.0395903587341309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,1,power_law_1.2,0.48801279067993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,1,power_law_1.2,1.1738368034362794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,1,power_law_1.01,6.57391357421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,1,power_law_1.2,1.3079296112060548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,1,power_law_1.01,0.5703551769256592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,1,power_law_1.2,1.62674560546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,1,power_law_1.01,0.6680255889892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,1,power_law_1.2,0.5011903762817382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,1,power_law_1.01,0.8721152305603027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,1,power_law_1.2,1.9412160873413087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,1,power_law_1.2,0.5379392147064209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,1,power_law_1.01,1.0815872192382812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,1,power_law_1.2,0.5638271808624268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,1,power_law_1.01,8.618701171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,1,power_law_1.2,2.601446342468262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,1,power_law_1.01,1.4878911972045898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,1,power_law_1.01,1.902822494506836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,1,power_law_1.2,3.34716796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,1,power_law_1.2,0.060812801122665405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,1,power_law_1.2,0.10069760084152221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,1,power_law_1.2,0.15261440277099608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,1,power_law_1.2,0.6242752075195312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,1,power_law_1.2,0.21059200763702393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,1,power_law_1.2,0.6238656044006348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,1,power_law_1.2,0.2759360074996948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,1,power_law_1.2,0.6219071865081787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,1,power_law_1.2,0.4592448234558105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,1,power_law_1.2,0.6836544036865234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,1,power_law_1.2,0.4806528091430664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,1,power_law_1.2,0.4898816108703613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,1,power_law_1.2,0.7045504093170166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,1,power_law_1.2,0.5070720195770264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,1,power_law_1.2,0.7585855960845947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,1,power_law_1.2,0.5033919811248779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,1,power_law_1.2,0.8359295845031738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,1,power_law_1.2,0.5166207790374756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,1,power_law_1.2,1.0030207633972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,1,power_law_1.2,0.5262784004211426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,1,power_law_1.2,0.547212791442871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,1,power_law_1.2,1.1565183639526366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,1,power_law_1.2,0.034892800450325015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,1,power_law_1.2,0.05422719717025757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,1,power_law_1.2,0.5818816184997558
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,1,power_law_1.2,0.07313920259475708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,1,power_law_1.2,1.4804672241210937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,1,power_law_1.2,0.6083392143249512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,1,power_law_1.2,0.09459199905395507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,1,power_law_1.2,0.6347839832305908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,1,power_law_1.2,0.12243200540542602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,1,power_law_1.2,1.8222719192504884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,1,power_law_1.2,0.7042687892913818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,1,power_law_1.2,0.8158975601196289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,1,power_law_1.2,0.9498368263244629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,1,power_law_1.2,2.500876808166504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,1,power_law_1.2,1.201516819000244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,1,power_law_1.2,0.19330559968948363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,1,power_law_1.2,1.4622591972351073
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,1,power_law_1.2,0.2041088104248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,1,power_law_1.2,0.9041728019714356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,1,power_law_1.2,3.1541120529174806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,1,balanced,0.14657599727312723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,1,balanced,0.32062933842341107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,1,balanced,0.5259679953257242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,1,power_law_1.2,1.956537628173828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,1,balanced,1.0016693274180095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,1,balanced,1.956266721089681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,1,power_law_1.2,4.505849456787109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,1,balanced,2.429114659627279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,1,power_law_1.2,2.4794687271118163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,1,balanced,2.4538826942443848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,1,power_law_1.2,0.5969535827636718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,1,balanced,2.4519359270731607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,1,power_law_1.2,0.20997118949890137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,1,power_law_1.2,4.711270523071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,1,power_law_1.2,0.21729280948638915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,1,balanced,2.4555892944335938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,1,power_law_1.2,0.2258687973022461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,1,power_law_1.2,0.2400063991546631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,1,balanced,2.4856106440226235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,1,power_law_1.2,0.25430400371551515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,1,power_law_1.2,0.2682431936264038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,1,power_law_1.2,3.5063743591308594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,1,balanced,2.5166613260904946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,1,power_law_1.2,0.3073983907699585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,1,power_law_1.2,0.2691391944885254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,1,balanced,2.5141493479410806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,1,power_law_1.2,6.400621032714843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,1,power_law_1.2,0.691974401473999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,1,balanced,2.747626622517904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,1,balanced,2.7796319325764975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,1,power_law_1.2,0.28945920467376707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,1,balanced,2.784501393636068
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,1,power_law_1.2,0.32997119426727295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,1,power_law_1.2,0.33356800079345705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,1,power_law_1.2,0.38327679634094236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,1,balanced,2.8208373387654624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,1,power_law_1.2,0.39347200393676757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,1,power_law_1.2,3.3070655822753907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,1,power_law_1.2,0.4515520095825195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,1,balanced,2.856821378072103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,1,balanced,0.12010666728019714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,1,power_law_1.2,2.979257583618164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,1,balanced,2.96997865041097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,1,balanced,3.130799929300944
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,1,power_law_1.2,0.5239935874938965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,1,balanced,0.19542932510375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,1,balanced,0.3449759880701701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,1,balanced,0.6529333194096884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,1,balanced,1.2479253609975178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,1,balanced,1.55021333694458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,1,balanced,0.13428266843159994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,1,balanced,1.5560426712036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,1,balanced,1.5608426729838054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,1,balanced,4.7961225509643555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,1,balanced,1.561829408009847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,1,balanced,1.5672426223754883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,1,power_law_1.2,4.5315711975097654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,1,balanced,1.5747092564900715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,1,balanced,4.660890579223633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,1,balanced,0.2015626629193624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,1,balanced,0.3469760020573934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,1,balanced,1.6298400561014812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,1,balanced,1.6339200337727864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,1,balanced,0.6297546625137329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,1,balanced,1.19268798828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,1,power_law_1.2,0.8887295722961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,1,power_law_1.2,0.6523392200469971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,1,balanced,1.4754346211751301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,1,balanced,7.388821283976237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,1,balanced,1.4846347173055012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,1,balanced,1.49018128712972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,1,balanced,1.4976320266723633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,1,balanced,1.6508639653523762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,1,balanced,1.5051573117574055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,1,power_law_1.2,0.8340991973876953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,1,balanced,1.671738624572754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,1,balanced,1.5175466537475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,1,balanced,8.988677342732748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,1,balanced,1.5300854047139485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,1,power_law_1.2,5.849017715454101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,1,balanced,1.5430986086527507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,1,balanced,1.5656800270080566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,1,balanced,1.5905866622924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,1,balanced,1.6137919425964355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,1,balanced,1.728373368581136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,1,balanced,1.6556746164957683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,1,power_law_1.2,1.1025216102600097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,1,balanced,1.8040053049723308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,1,balanced,1.7334507306416829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,1,balanced,12.809813181559244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,1,balanced,1.9818399747212727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,1,balanced,1.8179252942403157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,1,balanced,2.1512907346089682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,1,balanced,2.8136959075927734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,1,balanced,3.4590934117635093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,1,balanced,2.9685068130493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,1,balanced,3.766127904256185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,1,balanced,4.321050643920898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,1,balanced,5.495973587036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,1,balanced,5.557029088338216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,1,balanced,18.579984029134113
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,1,power_law_1.2,1.0499584197998046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,1,balanced,7.161210378011067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,1,balanced,7.227365493774414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,1,balanced,9.718026479085287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,1,power_law_1.2,6.580960083007812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,1,balanced,9.712703704833984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,1,balanced,13.090347290039062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,1,balanced,28.7925287882487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,1,balanced,14.070181528727213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,1,balanced,18.90620295206706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,1,power_law_1.2,8.538259124755859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,1,balanced,18.070053100585938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,1,balanced,39.81859842936198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,1,power_law_1.01,0.14555519819259644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,1,power_law_1.01,0.3187455892562866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,1,power_law_1.01,0.5245696067810058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,1,balanced,27.010208129882812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,1,power_law_1.01,0.836524772644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,1,power_law_1.01,0.11743359565734864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,1,power_law_1.01,1.15664644241333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,1,power_law_1.01,0.19278719425201415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,1,power_law_1.01,0.343340802192688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,1,power_law_1.01,1.9211776733398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,1,balanced,0.062037333846092224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,1,balanced,0.09070400396982829
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,1,balanced,0.14653866489728293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,1,balanced,0.2640533248583476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,1,balanced,0.4922613302866618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,1,power_law_1.01,2.062073516845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,1,balanced,0.6114559968312582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,1,power_law_1.2,9.325350189208985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,1,balanced,0.6142133474349976
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,1,balanced,0.6209919850031534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,1,power_law_1.01,2.0896703720092775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,1,balanced,0.6247626543045044
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,1,balanced,0.6291626691818237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,1,power_law_1.01,0.5443903923034668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,1,power_law_1.01,2.183558464050293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,1,power_law_1.01,0.7673984050750733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,1,power_law_1.01,1.2242048263549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,1,power_law_1.01,2.2155584335327148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,1,power_law_1.01,1.2599743843078612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,1,balanced,0.6352320114771525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,1,power_law_1.01,2.358054351806641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,1,power_law_1.01,1.3323840141296386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,1,power_law_1.01,1.3606847763061523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,1,power_law_1.01,2.553299140930176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,1,power_law_1.01,1.479257583618164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,1,power_law_1.01,2.581145668029785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,1,power_law_1.01,1.627724838256836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,1,balanced,0.6440213521321615
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,1,balanced,0.6981173356374105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,1,balanced,0.7105226516723633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,1,power_law_1.01,1.7371135711669923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,1,power_law_1.01,2.607289505004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,1,balanced,0.7245813210805258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,1,balanced,0.7147520383199056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,1,power_law_1.01,1.514521598815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,1,balanced,0.7324000199635824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,1,balanced,0.7999467055002848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,1,power_law_1.2,8.62635498046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,1,power_law_1.01,2.7584447860717773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,1,balanced,0.8265706698099772
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,1,balanced,1.2879467010498047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,1,balanced,1.0246293544769287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,1,power_law_1.01,2.7644224166870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,1,balanced,3.1954453786214194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,1,power_law_1.01,3.010585594177246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,1,power_law_1.01,1.5701824188232423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,1,balanced,1.9471680323282878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,1,power_law_1.01,1.6963327407836915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,1,balanced,2.8187732696533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,1,power_law_1.01,3.394144058227539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,1,power_law_1.01,1.7740287780761719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,1,power_law_1.01,0.13344639539718628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,1,balanced,3.4609972635904946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,1,power_law_1.01,0.19939199686050416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,1,power_law_1.01,0.34344959259033203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,1,power_law_1.01,1.9719039916992187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,1,power_law_1.01,3.91475830078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,1,power_law_1.01,0.55382399559021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,1,balanced,3.0714025497436523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,1,power_law_1.01,0.7170176029205322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,1,power_law_1.01,2.3244224548339845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,1,power_law_1.01,1.1954048156738282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,1,balanced,5.065082550048828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,1,power_law_1.01,4.552032089233398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,1,power_law_1.01,1.2188159942626953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,1,power_law_1.2,1.4683839797973632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,1,power_law_1.01,1.2679424285888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,1,power_law_1.01,1.342848014831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,1,power_law_1.01,1.3374015808105468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,1,power_law_1.01,5.430182266235351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,1,power_law_1.01,2.652979278564453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,1,power_law_1.01,1.3711615562438966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,1,power_law_1.01,1.425318431854248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,1,power_law_1.01,3.3711166381835938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,1,power_law_1.01,1.4326463699340821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,1,power_law_1.01,7.285298919677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,1,power_law_1.01,1.4869952201843262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,1,power_law_1.01,4.1696830749511715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,1,power_law_1.01,1.5384896278381348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,1,power_law_1.01,1.6060928344726562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,1,power_law_1.01,1.6973184585571288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,1,power_law_1.01,5.5976512908935545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,1,power_law_1.01,1.9785919189453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,1,power_law_1.01,9.367097473144531
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,1,power_law_1.01,0.06004480123519897
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,1,power_law_1.01,0.08925439715385437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,1,power_law_1.01,2.206195259094238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,1,power_law_1.01,0.14497920274734497
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,1,power_law_1.01,0.2218048095703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,1,power_law_1.01,0.3069567918777466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,1,power_law_1.01,7.105017852783203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,1,power_law_1.01,2.7326528549194338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,1,power_law_1.01,0.48448638916015624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,1,power_law_1.01,0.5106751918792725
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,1,power_law_1.01,0.5551360130310059
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,1,power_law_1.01,0.5630527973175049
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,1,power_law_1.01,0.5562943935394287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,1,power_law_1.01,0.6143487930297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,1,power_law_1.01,12.864659118652344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,1,power_law_1.01,0.6239424228668213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,1,power_law_1.01,0.6740352153778076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,1,power_law_1.01,10.138349151611328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,1,power_law_1.01,3.242860794067383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,1,power_law_1.01,0.7585536003112793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,1,power_law_1.01,0.8500479698181153
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,1,power_law_1.01,0.7283904075622558
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,1,power_law_1.01,0.8071999549865723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,1,power_law_1.01,4.320940780639648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,1,power_law_1.01,0.8081151962280273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,1,power_law_1.01,0.9159808158874512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,1,power_law_1.01,0.9429759979248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,1,power_law_1.01,5.383910369873047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,1,power_law_1.01,1.086348819732666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,1,power_law_1.01,16.656352233886718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,1,power_law_1.01,13.173715209960937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,1,power_law_1.01,1.2431103706359863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,1,power_law_1.01,1.471225643157959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,1,power_law_1.01,1.917068862915039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,1,power_law_1.2,1.5023743629455566
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,1,power_law_1.01,2.3495296478271483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,1,power_law_1.01,7.553107452392578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,1,power_law_1.01,3.2732158660888673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,1,power_law_1.2,11.244857788085938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,1,power_law_1.2,0.1450943946838379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,1,power_law_1.2,0.3168576002120972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,1,power_law_1.01,19.265049743652344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,1,power_law_1.01,5.178022384643555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,1,power_law_1.2,0.5193471908569336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,1,power_law_1.2,0.7658624172210693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,1,power_law_1.01,9.71146240234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,1,power_law_1.2,1.108409595489502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,1,power_law_1.2,1.8352319717407226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,1,power_law_1.2,12.435987091064453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,1,power_law_1.2,2.0357248306274416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,1,power_law_1.2,0.12031999826431275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,1,power_law_1.01,23.903077697753908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,1,power_law_1.2,0.19296640157699585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,1,power_law_1.2,2.183072090148926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,1,power_law_1.2,0.34428160190582274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,1,power_law_1.2,0.48772478103637695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,1,power_law_1.2,2.2096063613891603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,1,power_law_1.2,0.696895980834961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,1,power_law_1.2,1.1991616249084474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,1,power_law_1.2,2.2328960418701174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,1,power_law_1.01,13.965830993652343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,1,power_law_1.2,1.303916835784912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,1,power_law_1.2,0.1336127996444702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,1,power_law_1.2,2.404204750061035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,1,power_law_1.2,1.3346431732177735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,1,power_law_1.2,0.19861760139465331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,1,power_law_1.2,0.33886721134185793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,1,power_law_1.2,1.4435903549194335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,1,power_law_1.2,2.6359615325927734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,1,power_law_1.2,1.5017408370971679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,1,power_law_1.01,25.366514587402342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,1,power_law_1.2,2.604070472717285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,1,power_law_1.2,1.662259292602539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,1,power_law_1.2,1.8257535934448241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,1,power_law_1.2,2.6785280227661135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,1,power_law_1.2,0.48732800483703614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,1,power_law_1.2,0.6571968078613282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,1,power_law_1.2,1.1496512413024902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,1,power_law_1.2,1.2406271934509276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,1,power_law_1.2,1.2969856262207031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,1,power_law_1.01,18.213075256347658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,1,power_law_1.2,1.5792320251464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,1,power_law_1.2,1.3156224250793458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,1,power_law_1.2,2.788492774963379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,1,power_law_1.2,1.7016639709472656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,1,power_law_1.2,1.355187225341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,1,power_law_1.2,1.7524927139282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,1,power_law_1.2,1.3978303909301757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,1,power_law_1.2,2.841913604736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,1,power_law_1.2,1.4536640167236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,1,power_law_1.2,1.8435840606689453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,1,balanced,0.1495466629664103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,1,power_law_1.2,1.466329574584961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,1,balanced,0.29123733441034955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,1,power_law_1.01,31.382040405273436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,1,power_law_1.2,1.9823616027832032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,1,balanced,0.5136906703313192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,1,balanced,0.9429546991984049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,1,power_law_1.2,1.5332672119140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,1,balanced,1.8393120765686035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,1,power_law_1.2,2.3793664932250977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,1,power_law_1.2,1.6121343612670898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,1,power_law_1.2,0.05904639959335327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,1,balanced,0.11106666922569275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,1,balanced,0.1919040083885193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,1,power_law_1.2,3.0511167526245115
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,1,power_law_1.2,0.08748800158500672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,1,balanced,0.32310932874679565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,1,power_law_1.2,1.653785514831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,1,balanced,0.604965329170227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,1,power_law_1.2,0.14298880100250244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,1,power_law_1.2,2.724934387207031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,1,balanced,1.1780426502227783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,1,power_law_1.2,1.7646015167236329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,1,power_law_1.2,3.383967971801758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,1,balanced,2.2997066179911294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,1,power_law_1.2,2.0305471420288086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,1,balanced,3.4403467178344727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,1,balanced,3.615274747212728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,1,power_law_1.2,3.826163101196289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,1,power_law_1.2,2.2226240158081056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,1,power_law_1.2,0.21018879413604735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,1,power_law_1.2,0.2727679967880249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,1,balanced,3.5213279724121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,1,power_law_1.2,0.4689343929290771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,1,power_law_1.2,2.7591808319091795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,1,balanced,5.385546366373698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,1,power_law_1.2,3.4380352020263674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,1,power_law_1.2,0.5175104141235352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,1,balanced,3.4758825302124023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,1,power_law_1.2,4.632767868041992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,1,power_law_1.2,3.272230529785156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,1,balanced,3.5093866984049478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,1,balanced,5.514746983846028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,1,power_law_1.2,4.190502548217774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,1,balanced,3.5356480280558267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,1,power_law_1.2,5.454694366455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,1,power_law_1.2,0.5349120140075684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,1,power_law_1.2,4.359097671508789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,1,balanced,5.5586293538411455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,1,power_law_1.2,0.5599359989166259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,1,balanced,3.5615466435750327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,1,power_law_1.2,0.588102388381958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,1,power_law_1.2,0.6062911987304688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,1,power_law_1.2,5.699558258056641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,1,balanced,3.572842597961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,1,power_law_1.2,0.6542528152465821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,1,balanced,5.596591949462891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,1,power_law_1.2,5.410559844970703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,1,power_law_1.2,0.6825471878051758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,1,power_law_1.2,7.626464080810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,1,balanced,3.585754712422689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,1,balanced,5.475973129272461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,1,balanced,3.720826784769694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,1,power_law_1.2,7.196383666992188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,1,balanced,3.8821385701497397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,1,power_law_1.2,7.559308624267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,1,balanced,5.816373189290364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,1,power_law_1.2,9.342668914794922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,1,balanced,3.8892745971679688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,1,balanced,5.801247914632161
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,1,power_law_1.2,0.794214391708374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,1,power_law_1.2,10.226630401611327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,1,power_law_1.2,0.8852031707763672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,1,power_law_1.2,0.7573823928833008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,1,balanced,5.812469482421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,1,power_law_1.2,0.851411247253418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,1,power_law_1.2,0.8447104454040527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,1,balanced,4.069461186726888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,1,power_law_1.2,0.9542143821716309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,1,balanced,5.847653071085612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,1,power_law_1.2,0.9873920440673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,1,balanced,4.0837812423706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,1,power_law_1.2,12.916319274902344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,1,power_law_1.2,1.120140838623047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,1,balanced,5.730597178141276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,1,balanced,4.3712161382039385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,1,power_law_1.2,13.259834289550781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,1,balanced,4.767152150472005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,1,balanced,6.056383768717448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,1,power_law_1.2,1.3166144371032715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,1,balanced,5.381146748860677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,1,power_law_1.2,1.5263872146606445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,1,balanced,6.1417388916015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,1,power_law_1.2,1.9814655303955078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,1,power_law_1.2,1.9240192413330077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,1,power_law_1.2,16.824652099609374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,1,balanced,6.2550614674886065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,1,balanced,8.222298940022787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,1,power_law_1.2,9.664492797851562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,1,balanced,6.586245218912761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,1,power_law_1.2,19.320166015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,1,balanced,9.75050163269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,1,power_law_1.2,2.4098112106323244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,1,balanced,6.857290903727214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,1,balanced,7.507151921590169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,1,balanced,13.40878423055013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,1,balanced,0.10341333349545796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,1,balanced,0.17657599846522012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,1,balanced,0.3216266632080078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,1,balanced,0.5734933217366537
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,1,power_law_1.2,3.3234752655029296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,1,balanced,1.1068906784057617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,1,balanced,11.474683125813803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,1,balanced,2.1628106435139975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,1,balanced,3.2505973180135093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,1,balanced,3.260469436645508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,1,balanced,18.19745127360026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,1,balanced,3.260202725728353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,1,balanced,12.186912536621094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,1,balanced,3.2756481170654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,1,balanced,3.2837546666463218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,1,power_law_1.2,25.43402862548828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,1,balanced,3.290250778198242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,1,balanced,3.3010079065958657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,1,balanced,3.336666742960612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,1,power_law_1.2,13.899494934082032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,1,balanced,18.036309560139973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,1,balanced,3.4012959798177085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,1,balanced,25.51299285888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,1,balanced,3.438506762186686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,1,power_law_1.01,0.1454975962638855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,1,balanced,3.4872105916341147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,1,balanced,3.556426684061686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,1,power_law_1.2,2.300569534301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,1,balanced,3.6789652506510415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,1,power_law_1.01,0.29651200771331787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,1,power_law_1.01,0.525055980682373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,1,balanced,3.825381278991699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,1,power_law_1.01,0.951251220703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,1,power_law_1.01,0.1105728030204773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,1,power_law_1.01,0.19199999570846557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,1,balanced,25.986549377441406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,1,balanced,4.009423891703288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,1,power_law_1.01,1.6686080932617187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,1,power_law_1.01,0.32162559032440186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,1,power_law_1.01,0.6055744171142579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,1,power_law_1.01,2.344601631164551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,1,balanced,4.553104082743327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,1,power_law_1.01,1.0729599952697755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,1,power_law_1.01,2.7168703079223633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,1,power_law_1.01,1.4980863571166991
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,1,power_law_1.2,5.414969635009766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,1,balanced,0.05830933153629303
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,1,balanced,0.08629866441090901
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,1,balanced,0.14018666744232178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,1,balanced,0.24836800495783487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,1,power_law_1.01,1.7730560302734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,1,balanced,0.46393601099650067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,1,balanced,0.8928799629211426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,1,power_law_1.01,3.6627071380615233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,1,balanced,1.330949306488037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,1,power_law_1.01,2.315091133117676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,1,balanced,1.3300159772237141
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,1,balanced,1.333994706471761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,1,balanced,6.583893458048503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,1,power_law_1.01,3.8051326751708983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,1,balanced,1.336389382680257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,1,power_law_1.01,2.3482303619384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,1,balanced,1.3462986946105957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,1,power_law_1.2,18.141754150390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,1,balanced,1.3540852864583333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,1,balanced,1.3646880785624187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,1,power_law_1.01,3.8817279815673826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,1,power_law_1.2,24.69767608642578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,1,balanced,7.313413619995117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,1,balanced,1.3789706230163574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,1,balanced,1.3963039716084797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,1,balanced,1.4130345980326335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,1,balanced,1.5424532890319824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,1,power_law_1.01,4.154348754882813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,1,power_law_1.01,2.512428855895996
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,1,balanced,1.60315736134847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,1,balanced,1.5984373092651367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,1,balanced,9.997589111328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,1,balanced,1.6857280731201172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,1,power_law_1.01,4.250611114501953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,1,balanced,1.8109173774719238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,1,balanced,42.95518493652344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,1,power_law_1.01,0.10151679515838623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,1,balanced,1.920464038848877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,1,power_law_1.01,0.17577600479125977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,1,power_law_1.01,0.31856000423431396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,1,power_law_1.2,0.1478976011276245
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,1,balanced,2.8534507751464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,1,power_law_1.01,4.47644157409668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,1,power_law_1.01,2.6012863159179687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,1,power_law_1.2,0.2862911939620972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,1,power_law_1.01,0.5712895870208741
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,1,balanced,2.3982186317443848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,1,power_law_1.2,0.5104512214660645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,1,power_law_1.01,0.997043228149414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,1,balanced,13.307866414388021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,1,power_law_1.01,2.7161855697631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,1,power_law_1.2,0.9385663986206054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,1,power_law_1.01,4.773081588745117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,1,power_law_1.01,1.4850687980651855
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,1,balanced,5.848063786824544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,1,power_law_1.2,1.522067165374756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,1,power_law_1.01,2.859110450744629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,1,power_law_1.01,1.7500608444213868
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,1,balanced,3.201514561971029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,1,power_law_1.2,2.149286460876465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,1,power_law_1.01,3.091756820678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,1,power_law_1.01,5.2707775115966795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,1,power_law_1.01,2.194963264465332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,1,power_law_1.01,0.05720319747924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,1,power_law_1.2,2.4929792404174806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,1,balanced,5.820117314656575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,1,power_law_1.01,2.270956802368164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,1,power_law_1.01,3.3907264709472655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,1,power_law_1.2,32.93177490234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,1,power_law_1.01,5.732531356811523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,1,power_law_1.2,3.568307113647461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,1,power_law_1.01,2.390060806274414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,1,balanced,19.55237325032552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,1,power_law_1.01,3.6486080169677733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,1,power_law_1.2,0.11099519729614257
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,1,power_law_1.01,0.0840511977672577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,1,power_law_1.2,0.19061119556427003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,1,power_law_1.01,2.4717056274414064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,1,power_law_1.2,0.32125439643859866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,1,power_law_1.2,3.6929664611816406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,1,power_law_1.01,5.0601856231689455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,1,power_law_1.01,3.2691455841064454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,1,power_law_1.2,0.6028223991394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,1,power_law_1.01,2.5323392868041994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,1,power_law_1.2,0.9704128265380859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,1,power_law_1.2,3.8548030853271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,1,power_law_1.01,3.6220096588134765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,1,power_law_1.01,0.13788800239562987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,1,power_law_1.2,1.308191967010498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,1,power_law_1.01,2.5913280487060546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,1,power_law_1.01,5.549164962768555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,1,power_law_1.01,0.24615039825439453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,1,power_law_1.2,1.6426431655883789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,1,power_law_1.01,0.41560959815979004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,1,power_law_1.01,2.7062847137451174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,1,power_law_1.2,3.939622497558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,1,power_law_1.01,4.054240036010742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,1,power_law_1.2,2.216556739807129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,1,power_law_1.01,2.79180793762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,1,power_law_1.01,5.982201766967774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,1,power_law_1.2,4.306943893432617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,1,power_law_1.2,2.3079872131347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,1,power_law_1.01,4.675526428222656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,1,power_law_1.01,2.8316287994384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,1,power_law_1.2,2.4250879287719727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,1,power_law_1.2,0.10177279710769653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,1,power_law_1.2,4.49431037902832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,1,power_law_1.01,2.9446079254150392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,1,power_law_1.01,6.787763214111328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,1,power_law_1.2,0.1752575993537903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,1,power_law_1.01,5.4228862762451175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,1,power_law_1.2,2.5488704681396483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,1,power_law_1.2,0.3191744089126587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,1,power_law_1.01,0.5982783794403076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,1,power_law_1.2,0.5697792053222657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,1,power_law_1.01,3.1911552429199217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,1,power_law_1.01,0.7295360088348388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,1,power_law_1.2,2.6406784057617188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,1,power_law_1.2,4.808396911621093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,1,power_law_1.2,0.9135040283203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,1,power_law_1.01,0.9265664100646973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,1,power_law_1.01,0.9480832099914551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,1,power_law_1.01,7.668550109863281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,1,power_law_1.2,2.8543167114257812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,1,power_law_1.01,0.9787391662597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,1,power_law_1.2,5.4160511016845705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,1,power_law_1.01,1.0384320259094237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,1,power_law_1.2,3.138310432434082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,1,power_law_1.2,0.05738880038261414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,1,power_law_1.01,1.0518848419189453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,1,power_law_1.2,0.08427519798278808
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,1,power_law_1.2,0.13880319595336915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,1,power_law_1.01,1.0979711532592773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,1,power_law_1.2,3.466105651855469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,1,power_law_1.2,0.24680960178375244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,1,power_law_1.2,5.7335552215576175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,1,power_law_1.01,9.2504638671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,1,power_law_1.01,1.1874752044677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,1,power_law_1.2,0.3806207895278931
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,1,power_law_1.01,1.2254655838012696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,1,power_law_1.2,0.548633623123169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,1,power_law_1.01,3.430828857421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,1,power_law_1.2,3.778905487060547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,1,power_law_1.01,1.3050047874450683
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,1,power_law_1.2,0.678169584274292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,1,power_law_1.2,5.264748764038086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,1,power_law_1.2,0.8883008003234864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,1,power_law_1.01,3.8444671630859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,1,power_law_1.2,0.9257408142089844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,1,power_law_1.01,6.872889709472656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,1,power_law_1.2,0.9656512260437011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,1,power_law_1.01,4.360006332397461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,1,power_law_1.2,5.678694534301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,1,power_law_1.01,1.464345645904541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,1,power_law_1.2,1.0225088119506835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,1,power_law_1.2,3.427110290527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,1,power_law_1.2,1.267251205444336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,1,power_law_1.01,1.8077247619628907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,1,power_law_1.01,11.255142211914062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,1,power_law_1.01,8.22234878540039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,1,power_law_1.2,3.7545665740966796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,1,power_law_1.01,5.444217681884766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,1,power_law_1.2,6.1847999572753904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,1,power_law_1.2,1.0596799850463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,1,power_law_1.2,4.1270912170410154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,1,power_law_1.2,1.1123519897460938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,1,power_law_1.2,1.5366016387939454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,1,power_law_1.01,6.399871826171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,1,power_law_1.2,1.1726847648620606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,1,power_law_1.2,7.08826904296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,1,power_law_1.2,2.128121566772461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,1,power_law_1.2,1.2524224281311036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,1,power_law_1.01,11.098957061767578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,1,power_law_1.2,1.325984001159668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,1,power_law_1.01,1.587052822113037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,1,power_law_1.2,2.199443244934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,1,power_law_1.01,14.34229736328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,1,power_law_1.2,1.536524772644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,1,power_law_1.01,1.9633920669555665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,1,power_law_1.01,8.473971557617187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,1,power_law_1.2,2.263315200805664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,1,power_law_1.2,4.903251266479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,1,power_law_1.2,7.879641723632813
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,1,power_law_1.2,1.8548736572265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,1,power_law_1.01,1.8144512176513672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,1,power_law_1.2,1.6633087158203126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,1,power_law_1.01,2.2615999221801757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,1,power_law_1.2,5.6295616149902346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,1,power_law_1.01,2.207379150390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,1,power_law_1.01,13.963917541503907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,1,power_law_1.2,2.481702423095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,1,power_law_1.2,9.823808288574218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,1,power_law_1.01,10.629977416992187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,1,power_law_1.01,2.770560073852539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,1,power_law_1.2,2.067647933959961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,1,power_law_1.2,2.513702392578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,1,power_law_1.2,7.035923004150391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,1,power_law_1.01,18.216429138183592
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,1,power_law_1.01,2.9494144439697267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,1,power_law_1.2,1.8959423065185548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,1,power_law_1.2,2.5732927322387695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,1,power_law_1.2,2.4058176040649415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,1,power_law_1.2,2.649465560913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,1,power_law_1.01,3.809651184082031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,1,power_law_1.2,2.318694305419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,1,power_law_1.2,11.696115112304687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,1,power_law_1.2,8.555987548828124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,1,power_law_1.2,2.7477951049804688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,1,power_law_1.2,2.8827327728271483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,1,power_law_1.01,14.742636108398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,1,power_law_1.2,2.7899776458740235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,1,power_law_1.01,6.049216079711914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,1,power_law_1.01,19.79003448486328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,1,balanced,0.06442666550477345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,1,balanced,0.09708266456921895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,1,balanced,0.18216532468795776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,1,balanced,0.32871466875076294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,1,balanced,0.6058026552200317
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,1,power_law_1.2,3.142220878601074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,1,balanced,1.165285348892212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,1,balanced,1.1694026788075764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,1,balanced,1.1690133412679036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,1,balanced,1.1728373368581135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,1,power_law_1.2,11.381926727294921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,1,balanced,1.174511988957723
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,1,power_law_1.2,4.133036804199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,1,balanced,1.1895253658294678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,1,power_law_1.2,14.9638916015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,1,balanced,1.1967840194702148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,1,power_law_1.01,24.794496154785158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,1,balanced,1.1952746709187825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,1,balanced,1.206986665725708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,1,balanced,1.3316960334777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,1,balanced,1.3414293924967449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,1,power_law_1.2,2.9992000579833986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,1,balanced,1.3617439270019531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,1,power_law_1.01,18.662725830078124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,1,power_law_1.2,7.013369750976563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,1,balanced,1.4034825960795085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,1,balanced,0.05579199890295664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,1,balanced,0.09155199925104777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,1,balanced,0.1363093356291453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,1,balanced,0.22733867168426514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,1,balanced,1.449994723002116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,1,balanced,0.4167626698811849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,1,balanced,0.7873546282450358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,1,balanced,1.5444159507751465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,1,balanced,0.7894079685211182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,1,power_law_1.2,14.23582763671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,1,balanced,0.792090654373169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,1,balanced,0.7889119784037272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,1,balanced,0.05707733333110809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,1,balanced,0.7903733253479004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,1,balanced,0.7951093514760336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,1,balanced,0.7979520161946615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,1,power_law_1.2,3.2146686553955077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,1,balanced,0.8006613254547119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,1,balanced,0.8087466557820638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,1,balanced,0.8249066670735677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,1,balanced,1.6667839686075847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,1,balanced,0.8340799808502197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,1,power_law_1.2,18.48948516845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,1,balanced,0.8629173437754313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,1,balanced,0.08853866656621297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,1,balanced,0.1295840044816335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,1,balanced,2.296506722768148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,1,balanced,0.9280052979787191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,1,balanced,0.9860106309254965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,1,power_law_1.01,25.49012451171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,1,balanced,2.640629291534424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,1,balanced,1.1288586457570393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,1,balanced,1.2728640238444011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,1,balanced,1.9467466672261555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,1,balanced,3.8403892517089844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,1,power_law_1.2,3.5145343780517577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,1,balanced,0.23107733329137167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,1,balanced,0.4161493380864461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,1,balanced,0.7708480358123779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,1,balanced,0.7760053475697836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,1,balanced,4.996309280395508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,1,balanced,0.7784159978230795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,1,balanced,0.7831520239512125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,1,balanced,0.7875040372212728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,1,power_law_1.01,31.73529052734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,1,power_law_1.2,19.986810302734376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,1,balanced,0.7940800189971924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,1,balanced,0.8000533580780029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,1,balanced,2.2604692776997886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,1,balanced,0.8064693609873453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,1,power_law_1.01,0.06543359756469727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,1,balanced,0.820746660232544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,1,power_law_1.01,0.09608960151672363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,1,balanced,7.671471913655599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,1,balanced,0.8351840178171793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,1,balanced,0.03141866624355316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,1,balanced,0.047509332497914634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,1,power_law_1.01,0.18115839958190919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,1,balanced,0.06717333197593689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,1,balanced,0.8461759885152181
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,1,balanced,0.10117333134015401
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,1,balanced,0.17174933354059854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,1,power_law_1.01,0.33457279205322266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,1,balanced,0.8695627053578695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,1,balanced,0.9186293284098307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,1,power_law_1.01,0.47013120651245116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,1,balanced,0.9637333552042643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,1,power_law_1.01,0.6233791828155517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,1,balanced,1.055029312769572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,1,power_law_1.01,0.8564928054809571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,1,balanced,1.2021973133087158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,1,balanced,3.2627414067586265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,1,power_law_1.2,4.0342144012451175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,1,power_law_1.2,25.3708740234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,1,power_law_1.01,0.8725440025329589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,1,balanced,1.7797279357910156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,1,balanced,0.31036800146102905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,1,balanced,10.655855814615885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,1,balanced,0.314026673634847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,1,balanced,0.3169599970181783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,1,power_law_1.01,0.9204287528991699
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,1,balanced,0.3182826638221741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,1,balanced,2.012986660003662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,1,balanced,0.32073599100112915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,1,balanced,4.273034731547038
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,1,balanced,0.32464534044265747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,1,power_law_1.01,0.9465727806091309
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,1,balanced,0.32842665910720825
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,1,balanced,0.3332853317260742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,1,balanced,0.3393760124842326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,1,power_law_1.2,4.426758575439453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,1,balanced,0.35788798332214355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,1,balanced,0.36538668473561603
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,1,balanced,0.3821440140406291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,1,balanced,0.38060800234476727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,1,balanced,0.4006613492965698
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,1,balanced,0.4374133348464966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,1,balanced,6.30618158976237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,1,balanced,0.46556798617045086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,1,balanced,0.6393013397852579
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,1,balanced,0.5587573448816935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,1,power_law_1.2,5.526822280883789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,1,balanced,2.858816146850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,1,power_law_1.01,0.9951359748840332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,1,power_law_1.01,1.0360832214355469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,1,power_law_1.01,0.055238401889801024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,1,power_law_1.2,25.790789794921874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,1,balanced,3.7173385620117188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,1,power_law_1.01,0.0883903980255127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,1,balanced,8.341072082519531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,1,power_law_1.01,1.0773056030273438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,1,power_law_1.01,0.13204480409622193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,1,power_law_1.01,1.164352035522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,1,power_law_1.01,0.22068479061126708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,1,balanced,1.4261172612508137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,1,power_law_1.2,6.4989570617675785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,1,power_law_1.01,1.197094440460205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,1,balanced,5.429365158081055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,1,power_law_1.01,1.2263872146606445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,1,power_law_1.01,0.05416960120201111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,1,power_law_1.01,0.08663039803504943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,1,power_law_1.01,1.2711551666259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,1,power_law_1.01,0.12770559787750244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,1,power_law_1.01,0.22800641059875487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,1,power_law_1.01,1.3874112129211427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,1,power_law_1.01,0.32325758934021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,1,balanced,0.8278400103251139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,1,power_law_1.01,0.32766079902648926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,1,power_law_1.01,0.4253119945526123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,1,balanced,1.1754773457845051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,1,power_law_1.01,1.4635647773742675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,1,power_law_1.01,0.4372288227081299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,1,power_law_1.2,8.61475830078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,1,power_law_1.01,0.5762944221496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,1,balanced,1.5302666028340657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,1,power_law_1.01,0.5708159923553466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,1,power_law_1.01,1.6573503494262696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,1,power_law_1.01,0.6139008045196533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,1,power_law_1.2,0.06104959845542908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,1,power_law_1.01,0.617900800704956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,1,power_law_1.2,0.09364479780197144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,1,power_law_1.01,0.6212224006652832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,1,power_law_1.01,1.9564416885375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,1,balanced,7.122389475504558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,1,power_law_1.2,32.216128540039065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,1,power_law_1.01,0.6429183959960938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,1,power_law_1.2,0.055327999591827395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,1,power_law_1.01,2.505567932128906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,1,power_law_1.2,0.08860160112380981
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,1,power_law_1.01,0.02874239981174469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,1,power_law_1.2,0.1325055956840515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,1,power_law_1.01,0.045542401075363156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,1,power_law_1.2,0.21984639167785644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,1,power_law_1.01,0.6230144023895263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,1,power_law_1.01,0.06318079829216003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,1,power_law_1.2,0.1779584050178528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,1,power_law_1.01,2.9423616409301756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,1,power_law_1.01,0.6268159866333007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,1,power_law_1.2,10.630553436279296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,1,power_law_1.2,0.30524160861968996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,1,power_law_1.01,0.6712255954742432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,1,power_law_1.2,0.054553598165512085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,1,power_law_1.2,0.4203711986541748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,1,power_law_1.2,0.08678399920463561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,1,power_law_1.01,0.7238080024719238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,1,power_law_1.2,0.12805119752883912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,1,power_law_1.01,4.0929920196533205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,1,power_law_1.01,0.7658624172210693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,1,power_law_1.2,0.21984639167785644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,1,power_law_1.01,0.8624320030212402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,1,power_law_1.2,0.2892224073410034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,1,power_law_1.01,0.09824640154838563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,1,power_law_1.01,0.7419904232025146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,1,power_law_1.2,0.3988287925720215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,1,power_law_1.01,0.14019839763641356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,1,power_law_1.01,0.6593023777008057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,1,power_law_1.01,0.7633279800415039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,1,power_law_1.01,0.17938560247421265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,1,power_law_1.2,0.5732096195220947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,1,power_law_1.01,0.6796031951904297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,1,power_law_1.01,5.330400085449218
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,1,power_law_1.01,0.23639039993286132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,1,power_law_1.01,0.8442239761352539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,1,power_law_1.2,0.5991040229797363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,1,power_law_1.01,0.6960639953613281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,1,power_law_1.01,0.2477247953414917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,1,power_law_1.01,0.9535872459411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,1,power_law_1.01,0.7318655967712402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,1,power_law_1.2,0.29247360229492186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,1,power_law_1.01,0.7558527946472168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,1,power_law_1.01,0.7791359901428223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,1,power_law_1.2,14.661158752441406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,1,power_law_1.2,0.5649856090545654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,1,power_law_1.01,7.816825866699219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,1,power_law_1.2,0.8500288009643555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,1,power_law_1.2,0.6220160007476807
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,1,power_law_1.01,0.26097280979156495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,1,power_law_1.2,0.642086410522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,1,power_law_1.2,0.3803776025772095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,1,power_law_1.2,0.6722112178802491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,1,power_law_1.2,0.5709824085235595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,1,power_law_1.2,0.6840384006500244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,1,power_law_1.01,0.8247488021850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,1,power_law_1.2,0.5889023780822754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,1,power_law_1.2,0.6916287899017334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,1,power_law_1.2,0.6080128192901612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,1,power_law_1.2,0.7606719970703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,1,power_law_1.2,0.639296007156372
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,1,power_law_1.01,0.2637311935424805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,1,power_law_1.2,0.8848959922790527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,1,power_law_1.2,0.7739136219024658
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,1,power_law_1.01,0.2736128091812134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,1,power_law_1.2,0.6916863918304443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,1,power_law_1.2,0.9115712165832519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,1,power_law_1.2,0.7893119812011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,1,power_law_1.01,10.28057632446289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,1,power_law_1.2,0.7423295974731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,1,power_law_1.2,0.8478143692016602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,1,power_law_1.2,0.8026880264282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,1,power_law_1.01,0.9115776062011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,1,power_law_1.2,0.9588479995727539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,1,power_law_1.2,0.901683235168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,1,power_law_1.01,1.0078144073486328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,1,power_law_1.01,1.0654720306396483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,1,power_law_1.2,1.058899211883545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,1,power_law_1.2,0.7654528141021728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,1,power_law_1.01,1.2142975807189942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,1,power_law_1.2,18.865945434570314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,1,power_law_1.01,1.3034367561340332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,1,power_law_1.2,0.7954368114471435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,1,power_law_1.2,1.2669759750366212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,1,power_law_1.2,0.9452223777770996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,1,power_law_1.2,0.8814208030700683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,1,power_law_1.01,1.4207743644714355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,1,power_law_1.01,1.5497535705566405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,1,power_law_1.2,1.4594880104064942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,1,power_law_1.2,0.9967552185058594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,1,power_law_1.2,1.0019071578979493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,1,power_law_1.01,1.8330175399780273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,1,power_law_1.2,1.0410176277160645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,1,balanced,0.06284800171852112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,1,balanced,0.09481599926948547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,1,power_law_1.2,1.875200080871582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,1,power_law_1.2,1.1263936042785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,1,power_law_1.01,0.2754431962966919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,1,power_law_1.2,1.0911040306091309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,1,power_law_1.01,2.2408191680908205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,1,power_law_1.01,0.28940160274505616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,1,power_law_1.2,1.3548224449157715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,1,power_law_1.2,1.19302396774292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,1,power_law_1.2,2.2920000076293947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,1,power_law_1.2,1.2047103881835937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,1,power_law_1.01,3.068601608276367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,1,power_law_1.01,2.044326400756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,1,power_law_1.2,1.2514623641967773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,1,balanced,0.18225600322087607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,1,power_law_1.2,3.1190464019775392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,1,balanced,0.33744001388549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,1,balanced,0.605189323425293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,1,power_law_1.2,1.307155227661133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,1,power_law_1.2,0.02959359884262085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,1,balanced,1.1705386638641357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,1,power_law_1.01,2.549555206298828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,1,power_law_1.01,0.3107264041900635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,1,power_law_1.2,0.04575999975204468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,1,power_law_1.01,3.8977024078369142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,1,power_law_1.2,1.4267328262329102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,1,power_law_1.2,0.06366080045700073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,1,power_law_1.2,1.6020864486694335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,1,power_law_1.2,3.9587329864501952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,1,power_law_1.2,0.09136639833450318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,1,power_law_1.2,1.5139904022216797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,1,power_law_1.2,0.12407040596008301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,1,power_law_1.01,3.547788619995117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,1,power_law_1.2,2.098944091796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,1,power_law_1.2,0.1737215995788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,1,power_law_1.2,1.7668159484863282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,1,power_law_1.2,0.2361664056777954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,1,power_law_1.01,5.547756958007812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,1,balanced,1.1751360098520915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,1,power_law_1.2,0.24584319591522216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,1,power_law_1.01,0.3318336009979248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,1,power_law_1.2,2.597177505493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,1,balanced,1.1790186564127605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,1,power_law_1.2,1.9899967193603516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,1,power_law_1.01,0.35952000617980956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,1,power_law_1.01,4.5734912872314455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,1,balanced,1.1825333436330159
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,1,power_law_1.01,0.4124800205230713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,1,balanced,1.1856746673583984
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,1,power_law_1.01,0.3928191900253296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,1,power_law_1.2,2.5294784545898437
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,1,power_law_1.01,0.45034241676330566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,1,power_law_1.2,3.616019058227539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,1,power_law_1.01,0.45060482025146487
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,1,power_law_1.01,0.5376768112182617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,1,power_law_1.2,5.617753601074218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,1,power_law_1.01,7.1855613708496096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,1,power_law_1.2,3.028678321838379
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,1,power_law_1.01,0.5466815948486328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,1,balanced,0.05444799860318502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,1,balanced,0.09053333600362141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,1,power_law_1.01,6.594303894042969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,1,balanced,0.13357866803805032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,1,balanced,0.22191466887791952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,1,power_law_1.01,0.6431039810180664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,1,balanced,0.4119253158569336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,1,power_law_1.2,4.6388286590576175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,1,balanced,0.7839146455128988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,1,power_law_1.01,0.8049920082092286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,1,balanced,0.7846293449401855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,1,balanced,0.787887970606486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,1,power_law_1.2,4.281158447265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,1,balanced,0.7877279917399088
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,1,power_law_1.01,0.9825216293334961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,1,power_law_1.2,0.25197439193725585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,1,balanced,0.7899306615193685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,1,balanced,0.7955093383789062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,1,power_law_1.2,0.25885438919067383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,1,power_law_1.2,7.2670654296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,1,power_law_1.01,1.3484416007995605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,1,balanced,0.7976106802622477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,1,power_law_1.2,0.2738624095916748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,1,balanced,0.8000053564707438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,1,power_law_1.2,0.27998719215393064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,1,balanced,0.8101600011189779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,1,power_law_1.2,6.665676879882812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,1,power_law_1.01,8.634099578857422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,1,power_law_1.2,0.2981951951980591
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,1,power_law_1.01,2.302195167541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,1,power_law_1.2,5.5347137451171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,1,power_law_1.2,0.3155711889266968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,1,power_law_1.01,0.06263039708137512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,1,balanced,0.05615466833114624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,1,balanced,0.0890880028406779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,1,balanced,0.1311253309249878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,1,balanced,0.23226133982340494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,1,balanced,0.41491198539733887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,1,balanced,0.7721760272979736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,1,balanced,1.1892000039418538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,1,balanced,0.7780373096466064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,1,balanced,0.8221279780069987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,1,balanced,1.197754700978597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,1,balanced,0.7805173397064209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,1,balanced,0.8337653477986654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,1,balanced,0.784495989481608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,1,balanced,1.2012053330739338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,1,balanced,0.8552107016245524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,1,balanced,0.7865920066833496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,1,balanced,1.2117813428243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,1,power_law_1.2,8.708512115478516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,1,balanced,0.7957653204600016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,1,power_law_1.01,0.09353600144386291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,1,balanced,0.80349334081014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,1,balanced,1.329423983891805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,1,power_law_1.01,0.1792512059211731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,1,balanced,0.8088053067525228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,1,power_law_1.01,0.055212801694869994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,1,balanced,1.3411572774251301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,1,balanced,0.8215200106302897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,1,power_law_1.01,0.3339711904525757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,1,power_law_1.01,0.08824319839477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,1,balanced,0.8338080247243246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,1,balanced,1.3595360120137532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,1,power_law_1.01,0.13217920064926147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,1,power_law_1.01,0.4835455894470215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,1,balanced,0.8461173375447592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,1,balanced,1.4017386436462402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,1,balanced,0.8717706998189291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,1,power_law_1.2,0.34225280284881593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,1,power_law_1.2,8.048006439208985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,1,balanced,0.9212480386098226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,1,balanced,1.4494345982869465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,1,power_law_1.2,0.3730240106582642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,1,balanced,0.9639733632405599
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,1,balanced,0.03137599925200144
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,1,balanced,0.046181331078211464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,1,power_law_1.2,0.4374335765838623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,1,balanced,0.06582933167616527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,1,balanced,1.0576106707255046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,1,power_law_1.2,0.41332478523254396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,1,balanced,1.2075786590576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,1,power_law_1.2,0.47261438369750974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,1,balanced,1.788922627766927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,1,power_law_1.01,0.220032000541687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,1,power_law_1.2,0.47839999198913574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,1,power_law_1.01,0.5994559764862061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,1,balanced,0.9269066651662191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,1,power_law_1.01,0.3148672103881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,1,balanced,2.026869297027588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,1,power_law_1.01,0.8622207641601562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,1,balanced,0.990442673365275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,1,power_law_1.01,0.40599679946899414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,1,balanced,1.5488640467325847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,1,balanced,0.10085866848627727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,1,balanced,0.17100266615549722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,1,power_law_1.01,0.9081727981567382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,1,balanced,2.8675254185994468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,1,power_law_1.01,0.574124813079834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,1,balanced,1.666975975036621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,1,power_law_1.2,10.205369567871093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,1,power_law_1.01,0.9382335662841796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,1,power_law_1.01,0.6094848155975342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,1,balanced,2.306543986002604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,1,power_law_1.01,0.6279104232788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,1,balanced,3.7380959192911782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,1,power_law_1.2,0.5602816104888916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,1,power_law_1.01,0.629369592666626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,1,balanced,2.67519474029541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,1,power_law_1.2,0.5700287818908691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,1,balanced,1.1285920143127441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,1,power_law_1.01,0.6777408123016357
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,1,balanced,0.30991466840108234
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,1,power_law_1.2,0.6694719791412354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,1,balanced,1.2707573572794597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,1,power_law_1.01,0.7271743774414062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,1,balanced,5.4525705973307295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,1,balanced,3.8617067337036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,1,power_law_1.2,0.8333503723144531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,1,balanced,1.9435572624206543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,1,power_law_1.01,0.7641791820526123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,1,power_law_1.01,0.05512319803237915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,1,power_law_1.01,0.9480640411376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,1,power_law_1.01,0.0875711977481842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,1,power_law_1.01,0.857759952545166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,1,balanced,2.2595094045003257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,1,power_law_1.01,0.9755392074584961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,1,power_law_1.01,0.12794239521026612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,1,power_law_1.01,0.733958387374878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,1,balanced,4.920154571533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,1,power_law_1.01,0.22766079902648925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,1,power_law_1.01,1.0141183853149414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,1,balanced,7.128255844116211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,1,balanced,3.257925351460775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,1,power_law_1.01,0.3121920108795166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,1,power_law_1.01,1.0830656051635743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,1,power_law_1.01,0.43050241470336914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,1,power_law_1.01,1.163161563873291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,1,power_law_1.01,0.5845568180084229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,1,power_law_1.2,1.0191231727600099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,1,balanced,4.268293380737305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,1,power_law_1.01,0.6124735832214355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,1,power_law_1.01,1.2004672050476075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,1,balanced,7.86839485168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,1,power_law_1.2,1.3587712287902831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,1,power_law_1.01,0.6328767776489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,1,balanced,0.31251732508341473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,1,power_law_1.01,1.22357120513916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,1,balanced,0.3155519962310791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,1,balanced,0.31707199414571124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,1,power_law_1.01,0.6459648132324218
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,1,balanced,0.32013867298762005
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,1,balanced,0.32418133815129596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,1,power_law_1.01,1.2766655921936034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,1,power_law_1.2,2.358515167236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,1,power_law_1.01,0.6741439819335937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,1,power_law_1.2,0.061977601051330565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,1,power_law_1.01,1.3910911560058594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,1,power_law_1.2,0.09388160109519958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,1,power_law_1.2,0.17807359695434571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,1,power_law_1.01,0.7634496212005615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,1,power_law_1.2,0.30847361087799074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,1,balanced,10.953530629475912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,1,power_law_1.01,0.8380672454833984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,1,power_law_1.2,0.42872958183288573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,1,balanced,6.297002792358398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,1,balanced,0.3279680013656616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,1,power_law_1.2,0.05514240264892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,1,power_law_1.01,0.9482624053955078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,1,power_law_1.2,0.5593408107757568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,1,power_law_1.2,0.09011840224266052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,1,power_law_1.01,0.6800511837005615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,1,power_law_1.2,0.13318400382995604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,1,power_law_1.2,0.8380800247192383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,1,power_law_1.01,0.6947328090667725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,1,power_law_1.2,0.2214656114578247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,1,power_law_1.01,1.4432448387145995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,1,power_law_1.2,0.8897343635559082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,1,power_law_1.2,0.2874752044677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,1,power_law_1.01,0.028992000222206115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,1,power_law_1.2,0.9453375816345215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,1,power_law_1.01,1.6910783767700195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,1,power_law_1.01,0.04567039906978607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,1,balanced,0.3323413332303365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,1,balanced,0.33854933579762775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,1,balanced,0.3575253486633301
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,1,balanced,0.3647306760152181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,1,power_law_1.01,1.0794943809509276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,1,balanced,0.38098665078481037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,1,balanced,0.3803306818008423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,1,balanced,0.3995199998219808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,1,power_law_1.01,1.3120512008666991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,1,power_law_1.01,0.7185919761657715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,1,power_law_1.2,0.3813119888305664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,1,power_law_1.01,0.7566207885742188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,1,balanced,8.326271692911783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,1,power_law_1.01,1.5452159881591796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,1,power_law_1.2,0.5671167850494385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,1,power_law_1.2,0.9483391761779785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,1,power_law_1.01,1.9306175231933593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,1,power_law_1.2,0.5966464042663574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,1,power_law_1.2,0.9887935638427734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,1,power_law_1.01,2.0442495346069336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,1,power_law_1.2,0.603116798400879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,1,power_law_1.2,1.0621376037597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,1,power_law_1.01,2.449600028991699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,1,power_law_1.2,1.1089088439941406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,1,power_law_1.01,0.06376320123672485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,1,power_law_1.2,1.2209280014038086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,1,power_law_1.01,0.09864959716796876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,1,power_law_1.01,2.944051170349121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,1,power_law_1.01,0.1351423978805542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,1,power_law_1.01,0.18061439990997313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,1,balanced,0.435973326365153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,1,power_law_1.2,0.6423999786376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,1,power_law_1.01,0.2405951976776123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,1,power_law_1.2,0.05392000079154968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,1,power_law_1.2,0.6917695999145508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,1,power_law_1.01,0.2542975902557373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,1,power_law_1.01,4.2935935974121096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,1,power_law_1.2,0.08759679794311523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,1,power_law_1.01,0.7661183834075928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,1,power_law_1.01,0.2580287933349609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,1,power_law_1.2,0.12807040214538573
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,1,power_law_1.01,0.2622208118438721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,1,power_law_1.01,0.8205056190490723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,1,power_law_1.2,0.21747839450836182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,1,power_law_1.2,1.218956756591797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,1,power_law_1.01,0.27343358993530276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,1,power_law_1.2,0.28784639835357667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,1,power_law_1.01,0.9079615592956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,1,power_law_1.01,0.28384640216827395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,1,power_law_1.2,1.2138303756713866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,1,power_law_1.2,0.3960256099700928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,1,power_law_1.01,0.29062399864196775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,1,power_law_1.01,2.5423168182373046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,1,power_law_1.01,5.363264083862305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,1,power_law_1.2,0.5708288192749024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,1,power_law_1.01,0.30773119926452636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,1,power_law_1.2,1.3296895980834962
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,1,power_law_1.01,0.3354367971420288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,1,power_law_1.2,0.7428607940673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,1,power_law_1.01,0.3589695930480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,1,power_law_1.2,0.7872128009796142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,1,power_law_1.01,0.4100800037384033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,1,power_law_1.01,3.5639232635498046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,1,power_law_1.2,0.8675392150878907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,1,power_law_1.01,0.3885312080383301
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,1,balanced,0.4636586507161458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,1,power_law_1.01,1.0186880111694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,1,power_law_1.01,0.44756479263305665
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,1,balanced,0.6382666826248169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,1,power_law_1.2,0.7619647979736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,1,balanced,0.5565866629282633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,1,power_law_1.01,0.46145920753479003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,1,power_law_1.2,0.8092864036560059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,1,power_law_1.01,1.2273216247558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,1,power_law_1.01,7.682176208496093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,1,power_law_1.2,0.604588794708252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,1,power_law_1.01,0.5363967895507813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,1,power_law_1.2,0.8643712043762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,1,power_law_1.01,4.554579162597657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,1,power_law_1.2,0.6173567771911621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,1,power_law_1.01,0.5423168182373047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,1,power_law_1.01,1.408556842803955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,1,power_law_1.2,1.0000063896179199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,1,power_law_1.2,0.6371583938598633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,1,power_law_1.01,0.6441279888153076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,1,power_law_1.2,0.6699903964996338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,1,power_law_1.2,1.1104063987731934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,1,power_law_1.01,0.8131903648376465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,1,power_law_1.2,0.6904831886291504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,1,balanced,1.4229920705159504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,1,power_law_1.01,0.9835391998291015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,1,power_law_1.2,0.6897920131683349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,1,power_law_1.2,1.3971136093139649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,1,power_law_1.2,0.7223104000091553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,1,power_law_1.01,6.6183616638183596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,1,power_law_1.01,1.338758373260498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,1,power_law_1.2,0.7547776222229003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,1,power_law_1.01,1.8278400421142578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,1,power_law_1.01,10.625907135009765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,1,power_law_1.2,0.7822271823883057
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,1,power_law_1.01,2.156550407409668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,1,power_law_1.2,0.8482239723205567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,1,power_law_1.2,1.3539839744567872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,1,power_law_1.01,2.247219276428223
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,1,balanced,0.8242346445719401
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,1,power_law_1.2,0.9475263595581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,1,balanced,1.1708746751149495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,1,power_law_1.2,1.0592703819274902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,1,power_law_1.2,1.5304384231567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,1,balanced,1.5266666412353516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,1,power_law_1.01,3.0779008865356445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,1,balanced,0.05194133520126343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,1,power_law_1.2,1.2610943794250489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,1,balanced,0.07084799806276958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,1,balanced,0.1086346705754598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,1,power_law_1.2,1.7331520080566407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,1,balanced,0.187226672967275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,1,balanced,0.32863465944925946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,1,balanced,0.43296531836191815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,1,power_law_1.2,1.4598464012145995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,1,balanced,0.4349493185679118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,1,balanced,0.4359626770019531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,1,power_law_1.2,1.9913600921630858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,1,balanced,0.43695465723673504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,1,power_law_1.2,1.5924480438232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,1,power_law_1.01,3.907904052734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,1,balanced,0.43910932540893555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,1,balanced,0.44119465351104736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,1,power_law_1.2,1.8625280380249023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,1,balanced,0.46779731909434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,1,balanced,0.47275733947753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,1,power_law_1.01,8.647602844238282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,1,power_law_1.2,2.1131967544555663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,1,balanced,0.48134398460388184
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,1,power_law_1.2,0.02855679988861084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,1,balanced,0.49127999941507977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,1,power_law_1.01,0.04656000137329101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,1,balanced,0.5018399953842163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,1,balanced,0.5243519941965739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,1,power_law_1.01,0.07274240255355835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,1,balanced,0.5720746517181396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,1,power_law_1.2,2.614937591552734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,1,balanced,0.6132959922154745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,1,power_law_1.01,5.556000137329102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,1,balanced,0.9822080135345459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,1,balanced,1.0570399761199951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,1,power_law_1.2,2.551545524597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,1,balanced,1.529813289642334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,1,power_law_1.2,3.626758575439453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,1,power_law_1.2,2.2928703308105467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,1,balanced,1.7447840372721355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,1,power_law_1.01,0.1363263964653015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,1,power_law_1.2,2.9853183746337892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,1,balanced,0.04860266546408335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,1,power_law_1.01,0.20563840866088867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,1,power_law_1.01,7.2012481689453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,1,power_law_1.2,3.106988716125488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,1,power_law_1.01,0.2771519899368286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,1,power_law_1.2,4.6403968811035154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,1,power_law_1.01,0.4467264175415039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,1,power_law_1.01,0.0473471999168396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,1,power_law_1.2,0.045561599731445315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,1,power_law_1.2,4.256025695800782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,1,power_law_1.2,0.06287999749183655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,1,power_law_1.2,3.9338623046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,1,balanced,2.6614452997843423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,1,balanced,3.359882672627767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,1,power_law_1.2,5.341081619262695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,1,power_law_1.01,0.472979211807251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,1,power_law_1.01,0.5033984184265137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,1,power_law_1.2,5.584159851074219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,1,power_law_1.2,0.09183359742164612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,1,power_law_1.01,0.49568638801574705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,1,power_law_1.2,0.12840319871902467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,1,power_law_1.01,0.5181695938110351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,1,balanced,4.981594721476237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,1,power_law_1.2,0.17056000232696533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,1,power_law_1.01,0.5361472129821777
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,1,power_law_1.2,0.23588480949401855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,1,power_law_1.01,0.5536064147949219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,1,power_law_1.2,6.669516754150391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,1,power_law_1.01,0.06802560091018676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,1,power_law_1.2,0.2474368095397949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,1,power_law_1.01,0.568345594406128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,1,power_law_1.01,0.10765440464019775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,1,power_law_1.2,7.767443084716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,1,power_law_1.01,0.16516480445861817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,1,power_law_1.01,0.5958720207214355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,1,power_law_1.01,0.2205631971359253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,1,balanced,6.60366948445638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,1,power_law_1.01,0.6419136047363281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,1,power_law_1.01,0.34201600551605227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,1,power_law_1.2,0.04632959961891174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,1,power_law_1.01,0.6649087905883789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,1,power_law_1.01,0.36583681106567384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,1,power_law_1.2,0.07363839745521546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,1,power_law_1.01,0.38246400356292726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,1,power_law_1.01,0.7122303962707519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,1,power_law_1.2,0.13765759468078614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,1,balanced,0.07542933523654938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,1,power_law_1.2,0.2510591983795166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,1,power_law_1.2,7.230873870849609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,1,power_law_1.2,8.714681243896484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,1,power_law_1.01,0.38705921173095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,1,power_law_1.2,0.194758403301239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,1,power_law_1.01,0.7193024158477783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,1,power_law_1.2,0.2607295989990234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,1,power_law_1.2,0.2583872079849243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,1,power_law_1.2,0.04880000054836273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,1,power_law_1.2,0.2701695919036865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,1,power_law_1.2,0.44262399673461916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,1,power_law_1.2,0.06715520024299622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,1,power_law_1.2,0.10708479881286621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,1,power_law_1.2,0.2837631940841675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,1,power_law_1.2,0.46674561500549316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,1,power_law_1.2,0.15571839809417726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,1,power_law_1.2,0.29573760032653806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,1,power_law_1.2,0.5028800010681153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,1,power_law_1.2,0.20385279655456542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,1,power_law_1.2,0.3184000015258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,1,power_law_1.01,0.39962239265441896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,1,power_law_1.2,0.5262335777282715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,1,power_law_1.2,0.33997440338134766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,1,power_law_1.2,0.3482752084732056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,1,power_law_1.01,0.4316671848297119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,1,power_law_1.2,0.3646399974822998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,1,power_law_1.2,0.37377281188964845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,1,power_law_1.2,0.5219840049743653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,1,power_law_1.01,0.44213118553161623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,1,power_law_1.2,0.37392640113830566
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,1,power_law_1.2,0.432096004486084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,1,power_law_1.01,0.4498112201690674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,1,power_law_1.2,0.3857599973678589
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,1,power_law_1.2,0.4056704044342041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,1,power_law_1.2,9.824102020263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,1,power_law_1.2,0.41001601219177247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,1,power_law_1.2,0.4396992206573486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,1,power_law_1.2,0.448364782333374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,1,power_law_1.2,0.4624512195587158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,1,power_law_1.2,0.48531198501586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,1,balanced,0.1896479924519857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,1,balanced,0.33314667145411175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,1,power_law_1.2,0.5135168075561524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,1,balanced,0.6183679898579916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,1,power_law_1.01,0.7826496124267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,1,power_law_1.01,0.46428799629211426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,1,balanced,0.13962666193644205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,1,power_law_1.01,0.48719358444213867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,1,power_law_1.01,0.8852095603942871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,1,power_law_1.01,0.5096191883087158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,1,power_law_1.01,0.9802623748779297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,1,power_law_1.01,0.5600959777832031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,1,power_law_1.01,0.648748779296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,1,power_law_1.01,1.193830394744873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,1,power_law_1.2,0.55349760055542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,1,power_law_1.01,0.7492224216461182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,1,balanced,0.28784533341725665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,1,power_law_1.2,0.582374382019043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,1,power_law_1.01,1.4228096008300781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,1,balanced,0.5125439961751302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,1,power_law_1.01,0.9366463661193848
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,1,power_law_1.2,0.4722623825073242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,1,power_law_1.2,0.6015744209289551
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,1,power_law_1.2,0.4766655921936035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,1,power_law_1.01,1.1364095687866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,1,power_law_1.2,0.6203839778900146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,1,power_law_1.01,1.879692840576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,1,power_law_1.2,0.5590015888214112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,1,power_law_1.2,0.645030403137207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,1,power_law_1.01,1.5333824157714844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,1,power_law_1.2,0.5698048114776612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,1,power_law_1.2,0.6977151870727539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,1,power_law_1.01,2.3553728103637694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,1,power_law_1.2,0.5356480121612549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,1,power_law_1.2,0.7371776103973389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,1,power_law_1.01,1.9238975524902344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,1,power_law_1.2,0.5814527988433837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,1,balanced,0.9621439774831136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,1,power_law_1.2,0.7596799850463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,1,power_law_1.2,0.6597248077392578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,1,power_law_1.01,3.2699710845947267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,1,power_law_1.2,0.7585599899291993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,1,power_law_1.01,2.737376022338867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,1,power_law_1.2,0.9572223663330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,1,power_law_1.2,0.6664639949798584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,1,balanced,1.2027520338694255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,1,power_law_1.2,1.1526335716247558
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,1,power_law_1.2,0.8471424102783203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,1,power_law_1.01,4.218489456176758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,1,power_law_1.01,3.5475391387939452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,1,power_law_1.2,1.5613311767578124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,1,power_law_1.2,0.8117952346801758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,1,power_law_1.2,1.960313606262207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,1,power_law_1.2,0.9175999641418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,1,balanced,1.8506293296813965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,1,power_law_1.2,1.0143487930297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,1,power_law_1.01,5.192979049682617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,1,power_law_1.2,2.7707839965820313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,1,power_law_1.2,1.227302360534668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,1,balanced,3.622213363647461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,1,power_law_1.2,1.0111680030822754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,1,balanced,0.24969599644343057
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,1,power_law_1.2,1.3611200332641602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,1,balanced,5.337952295939128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,1,power_law_1.01,6.83031005859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,1,balanced,2.330176035563151
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,1,power_law_1.2,2.3344640731811523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,1,power_law_1.2,1.4439295768737792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,1,balanced,3.410442670186361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,1,power_law_1.01,0.2835263967514038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,1,power_law_1.2,1.8860223770141602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,1,power_law_1.01,0.18637440204620362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,1,power_law_1.01,0.5046080112457275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,1,power_law_1.01,0.33036160469055176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,1,power_law_1.2,3.5809921264648437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,1,power_law_1.01,0.9642880439758301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,1,power_law_1.01,0.6183872222900391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,1,balanced,5.430517196655273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,1,power_law_1.2,2.330188751220703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,1,power_law_1.01,1.6654144287109376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,1,power_law_1.01,1.074009609222412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,1,balanced,0.1932906707127889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,1,balanced,0.3264159957567851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,1,balanced,3.4255574544270835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,1,balanced,0.5949920018513998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,1,balanced,1.1408960024515789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,1,balanced,5.348704020182292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,1,power_law_1.01,1.525011157989502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,1,power_law_1.2,3.265856170654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,1,balanced,3.4268480936686196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,1,balanced,2.195653279622396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,1,power_law_1.01,2.5476543426513674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,1,balanced,3.5416641235351562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,1,balanced,5.494112014770508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,1,balanced,0.43383467197418213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,1,balanced,0.5763733386993408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,1,balanced,3.5809653600056968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,1,power_law_1.01,2.3663360595703127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,1,power_law_1.01,2.729644775390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,1,balanced,5.552517573038737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,1,balanced,3.543365478515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,1,balanced,3.2366345723470054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,1,power_law_1.01,2.837881660461426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,1,power_law_1.01,3.9502079010009767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,1,balanced,3.7189013163248696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,1,balanced,5.597989400227864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,1,power_law_1.01,2.9312896728515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,1,balanced,3.7427892684936523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,1,power_law_1.01,4.346335983276367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,1,balanced,5.6310774485270185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,1,power_law_1.01,3.0802112579345704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,1,balanced,3.6891520818074546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,1,balanced,3.2358986536661782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,1,balanced,3.92083740234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,1,power_law_1.01,4.461414337158203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,1,power_law_1.01,3.392262268066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,1,power_law_1.2,5.20445442199707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,1,balanced,0.5781813462575277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,1,balanced,5.495450973510742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,1,power_law_1.01,4.552223968505859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,1,balanced,3.2526025772094727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,1,balanced,3.2710240681966147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,1,balanced,7.060522715250651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,1,power_law_1.01,4.67841911315918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,1,power_law_1.2,4.253855895996094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,1,balanced,3.3074560165405273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,1,balanced,5.748655954996745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,1,balanced,4.788634618123372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,1,balanced,3.3544158935546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,1,power_law_1.01,5.223782348632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,1,power_law_1.01,3.5997310638427735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,1,balanced,3.384357452392578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,1,balanced,5.125743865966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,1,balanced,3.4060160319010415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,1,power_law_1.01,3.852083206176758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,1,power_law_1.01,5.731622314453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,1,balanced,3.4587732950846353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,1,balanced,5.551631927490234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,1,balanced,8.257935841878256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,1,balanced,3.5600798924764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,1,power_law_1.01,4.459449768066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,1,balanced,5.625050862630208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,1,power_law_1.01,5.899039840698242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,1,balanced,3.606565475463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,1,balanced,8.919669469197592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,1,power_law_1.01,4.995846557617187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,1,balanced,3.721365292867025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,1,balanced,0.5838666756947836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,1,balanced,0.5851840178171793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,1,balanced,7.783941268920898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,1,power_law_1.2,6.847328186035156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,1,balanced,3.96996275583903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,1,power_law_1.01,6.865849304199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,1,power_law_1.01,5.608012771606445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,1,power_law_1.2,0.2846400022506714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,1,balanced,4.197018623352051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,1,power_law_1.2,0.508019208908081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,1,balanced,12.889503479003906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,1,balanced,9.866575876871744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,1,power_law_1.2,0.9638655662536622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,1,power_law_1.01,7.462624359130859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,1,balanced,6.582688013712565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,1,power_law_1.2,1.4913984298706056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,1,power_law_1.2,2.160710334777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,1,balanced,7.062906901041667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,1,power_law_1.01,6.720729827880859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,1,power_law_1.01,8.259334564208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,1,balanced,14.22213872273763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,1,power_law_1.2,3.839430236816406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,1,power_law_1.01,5.3068992614746096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,1,balanced,10.063658396402994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,1,balanced,0.5840533177057902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,1,power_law_1.2,4.24535026550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,1,balanced,6.47709846496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,1,power_law_1.01,0.18977919816970826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,1,power_law_1.2,4.394182586669922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,1,power_law_1.01,6.058553695678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,1,power_law_1.01,0.3218559980392456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,1,balanced,6.920906702677409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,1,power_law_1.01,0.5915840148925782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,1,balanced,10.813060760498047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,1,power_law_1.01,1.0366911888122559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,1,power_law_1.01,9.848281860351562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,1,power_law_1.2,4.650719833374024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,1,power_law_1.01,1.4351231575012207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,1,balanced,23.778345743815105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,1,power_law_1.01,7.523296356201172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,1,balanced,10.295706431070963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,1,power_law_1.01,2.43753604888916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,1,power_law_1.01,7.065497589111328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,1,power_law_1.01,2.6489536285400392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,1,balanced,16.697882334391277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,1,power_law_1.2,5.123871994018555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,1,balanced,11.46615473429362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,1,power_law_1.01,2.7780351638793945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,1,power_law_1.01,7.904819488525391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,1,power_law_1.01,2.8570175170898438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,1,power_law_1.2,5.258886337280273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,1,power_law_1.01,9.022994995117188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,1,power_law_1.01,2.891046333312988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,1,balanced,0.5901546478271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,1,balanced,0.5939253171284994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,1,balanced,0.6000800132751465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,1,balanced,0.6088533401489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,1,balanced,0.6162453492482504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,1,power_law_1.01,3.018783950805664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,1,power_law_1.2,5.787846374511719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,1,balanced,0.7282186349232992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,1,power_law_1.01,9.662477111816406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,1,balanced,20.337844848632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,1,balanced,18.17310969034831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,1,power_law_1.01,3.075641632080078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,1,balanced,32.35562642415365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,1,power_law_1.2,6.257638549804687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,1,power_law_1.01,12.03345947265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,1,power_law_1.01,3.1752832412719725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,1,power_law_1.01,3.2913856506347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,1,power_law_1.01,11.7615234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,1,power_law_1.2,7.094547271728516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,1,power_law_1.01,3.4210689544677733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,1,balanced,0.666693369547526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,1,balanced,20.68991978963216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,1,power_law_1.01,3.537094497680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,1,power_law_1.01,15.128851318359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,1,power_law_1.2,8.044556427001954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,1,power_law_1.01,3.854905700683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,1,balanced,29.866485595703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,1,power_law_1.01,15.078810119628907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,1,power_law_1.01,4.409056091308594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,1,power_law_1.2,8.694361877441406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,1,power_law_1.01,5.010047912597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,1,balanced,0.7078026930491129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,1,power_law_1.01,6.146239852905273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,1,power_law_1.2,10.152019500732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,1,balanced,50.12260437011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,1,power_law_1.01,21.202053833007813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,1,balanced,32.83974965413412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,1,power_law_1.01,18.66612548828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,1,power_law_1.01,7.305088043212891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,1,power_law_1.2,7.532006072998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,1,balanced,37.496500651041664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,1,balanced,0.7334667046864828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,1,power_law_1.2,8.172230529785157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,1,power_law_1.01,9.616556549072266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,1,power_law_1.2,0.18710399866104127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,1,power_law_1.2,0.33165440559387205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,1,power_law_1.2,0.6173120021820069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,1,power_law_1.2,0.9850560188293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,1,power_law_1.2,10.182994842529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,1,power_law_1.2,1.3692607879638672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,1,power_law_1.01,27.299200439453124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,1,power_law_1.01,11.933113861083985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,1,power_law_1.2,2.487436866760254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,1,balanced,41.2021230061849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,1,power_law_1.2,2.6794687271118165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,1,balanced,0.8285866578420004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,1,balanced,0.875274658203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,1,power_law_1.01,25.6406982421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,1,balanced,1.2297226587931316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,1,balanced,1.333621342976888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,1,power_law_1.2,2.853727912902832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,1,power_law_1.2,12.053548431396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,1,balanced,1.8910773595174153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,1,balanced,2.1082186698913574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,1,power_law_1.2,2.978688049316406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,1,balanced,3.224656105041504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,1,power_law_1.2,3.1569408416748046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,1,power_law_1.01,16.610105895996092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,1,balanced,3.986639976501465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,1,power_law_1.2,3.4859649658203127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,1,balanced,64.83476257324219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,2,balanced,0.16516799728075662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,2,balanced,0.22174400091171265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,2,balanced,0.30954132477442425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,2,balanced,0.4056053161621094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,2,balanced,0.4046186606089274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,1,power_law_1.2,15.539039611816406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,2,balanced,0.4079413414001465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,2,balanced,0.4094080130259196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,2,balanced,0.4117120107014974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,2,balanced,0.43088531494140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,2,balanced,0.4184906482696533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,2,balanced,0.42845332622528076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,2,balanced,0.4319466749827067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,2,balanced,0.44167999426523846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,2,balanced,0.4598720073699951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,2,balanced,0.5417653322219849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,2,balanced,0.5565653244654337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,1,power_law_1.2,3.759027099609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,1,power_law_1.2,4.050547027587891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,1,power_law_1.01,39.52451171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,2,balanced,0.5634026527404785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,2,balanced,0.764240026473999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,1,power_law_1.01,33.71782531738281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,2,balanced,0.8739519913991293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,2,balanced,1.3602506319681804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,1,power_law_1.2,4.672985458374024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,2,balanced,1.6813653310139973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,1,power_law_1.2,19.455751037597658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,2,balanced,2.7507092157999673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,1,power_law_1.01,21.208384704589843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,1,power_law_1.2,5.224531173706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,1,balanced,63.1742909749349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,2,balanced,3.7855361302693686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,1,power_law_1.2,5.826099014282226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,2,balanced,6.792997360229492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,1,power_law_1.2,6.867884826660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,2,balanced,9.06447982788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,1,power_law_1.2,0.18844159841537475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,1,power_law_1.2,26.782009887695313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,1,power_law_1.2,5.480780792236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,1,power_law_1.2,0.3221695899963379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,1,power_law_1.2,0.592736005783081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,1,power_law_1.2,0.9424832344055176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,1,power_law_1.2,6.19288330078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,1,power_law_1.2,1.3447999954223633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,2,balanced,14.889029184977213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,1,power_law_1.01,30.35203857421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,1,power_law_1.2,2.4036479949951173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,1,power_law_1.2,2.6220672607421873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,1,power_law_1.2,7.705484771728516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,1,power_law_1.2,2.757516860961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,1,power_law_1.01,49.790060424804686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,1,power_law_1.01,52.20460205078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,2,balanced,0.14568000038464865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,2,balanced,0.17659733692804971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,2,balanced,0.2267413338025411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,1,power_law_1.2,2.831302452087402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,2,balanced,0.29901333649953205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,2,balanced,0.30213866631189984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,2,balanced,0.3090026577313741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,2,balanced,0.31169599294662476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,2,balanced,0.32172266642252606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,2,balanced,0.32841066519419354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,2,balanced,20.82534917195638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,1,power_law_1.2,2.867136001586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,1,power_law_1.2,9.211219024658202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,2,balanced,0.12171733379364014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,2,balanced,0.15408000349998474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,2,balanced,0.2137440045674642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,2,balanced,0.21533334255218506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,2,balanced,0.21662932634353638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,2,balanced,0.2177120049794515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,2,balanced,0.21974400679270426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,1,power_law_1.2,3.0256128311157227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,2,balanced,0.22485866149266562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,2,balanced,0.22949333985646567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,2,balanced,0.2325813372929891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,2,balanced,0.23919467131296793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,2,balanced,0.2518133322397868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,2,balanced,0.33026667435963947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,2,balanced,0.25940799713134766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,2,balanced,0.3449546496073405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,2,balanced,0.29885333776474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,2,balanced,0.35473068555196124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,2,balanced,0.3450026512145996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,2,balanced,0.3715146780014038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,1,power_law_1.2,3.1324928283691404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,2,balanced,0.3977706829706828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,2,balanced,0.4776853322982788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,1,power_law_1.2,35.06869201660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,2,balanced,0.5015999873479208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,2,balanced,0.4994560082753499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,2,balanced,0.7672800223032633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,1,power_law_1.2,3.2863487243652343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,2,balanced,0.9282826582590739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,2,balanced,1.335792064666748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,1,power_law_1.2,12.235948944091797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,2,balanced,0.37431466579437256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,2,balanced,1.6969599723815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,1,power_law_1.2,3.3732608795166015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,2,balanced,2.5062506993611655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,2,balanced,3.3046932220458984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,1,power_law_1.2,3.5757312774658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,2,balanced,0.39927999178568524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,2,balanced,0.5698346694310507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,1,balanced,82.58828735351562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,2,balanced,0.6663093169530233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,1,power_law_1.01,39.29096984863281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,2,balanced,0.9549973011016846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,2,balanced,4.949573198954265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,2,power_law_1.01,0.20755200386047362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,1,power_law_1.2,3.708652877807617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,2,power_law_1.01,0.22899839878082276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,2,power_law_1.01,0.2682624101638794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,2,power_law_1.01,0.40176639556884763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,2,power_law_1.01,0.4141119956970215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,1,power_law_1.2,15.259724426269532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,2,balanced,6.999024073282878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,1,power_law_1.2,3.988505554199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,2,balanced,1.260101318359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,2,balanced,1.8431520462036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,2,balanced,2.3808266321818032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,1,power_law_1.2,4.538950347900391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,2,power_law_1.01,0.4245312213897705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,2,power_law_1.01,0.4098495960235596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,2,power_law_1.01,0.41258878707885743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,2,balanced,3.56055482228597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,2,balanced,12.264208475748697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,1,power_law_1.2,5.099948883056641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,2,balanced,4.627658526102702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,2,power_law_1.01,0.44019198417663574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,2,power_law_1.01,0.4438271999359131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,2,power_law_1.01,0.46462078094482423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,2,power_law_1.01,0.4705984115600586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,2,balanced,6.953493118286133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,1,power_law_1.2,6.2769214630126955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,2,power_law_1.01,0.4905407905578613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,2,power_law_1.01,0.5152383804321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,1,power_law_1.2,21.381639099121095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,2,power_law_1.01,0.60033278465271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,2,balanced,17.520821889241535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,2,power_law_1.01,0.14743679761886597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,2,power_law_1.01,0.6818431854248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,2,power_law_1.01,0.1970047950744629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,2,balanced,9.204970677693685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,2,power_law_1.01,0.1935871958732605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,1,power_law_1.2,7.431871795654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,2,power_law_1.01,0.29983360767364503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,2,power_law_1.01,0.3003328084945679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,2,power_law_1.01,0.12068480253219604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,2,power_law_1.01,0.7698560237884522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,1,power_law_1.01,66.89846801757812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,2,power_law_1.01,0.9869888305664063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,2,power_law_1.01,1.3280960083007813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,1,power_law_1.2,9.741331481933594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,2,power_law_1.01,0.1503551959991455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,2,power_law_1.01,0.1938367962837219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,2,power_law_1.01,0.21314558982849122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,2,power_law_1.01,0.20982398986816406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,2,balanced,0.04279466470082601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,2,power_law_1.01,0.21740798950195311
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,2,balanced,0.06021333237489065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,2,power_law_1.01,1.8518911361694337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,2,balanced,0.08371733625729878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,2,power_law_1.01,0.21982080936431886
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,2,balanced,0.08473599950472514
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,2,balanced,0.08683199683825175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,2,balanced,0.08854400118192036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,2,balanced,0.08896533648173015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,2,balanced,0.08957333366076152
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,2,balanced,0.09356799721717834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,2,balanced,0.09412266810735066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,2,power_law_1.01,0.3038655996322632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,2,power_law_1.01,0.3161344051361084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,1,power_law_1.2,27.528396606445312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,2,power_law_1.01,0.3263295888900757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,1,power_law_1.2,12.063539123535156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,2,power_law_1.01,0.330297589302063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,2,power_law_1.01,0.3450752019882202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,2,power_law_1.01,0.3580607891082764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,2,power_law_1.01,0.22371199131011962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,2,power_law_1.01,0.3997632026672363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,2,power_law_1.01,0.23189759254455566
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,2,balanced,0.095360000928243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,2,power_law_1.01,0.2381376028060913
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,2,balanced,0.12179733316103618
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,2,balanced,0.10975999633471172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,2,power_law_1.01,2.575494384765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,2,balanced,0.10355732838312785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,2,balanced,0.25917333364486694
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,2,balanced,0.18277867635091147
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,2,balanced,0.11306132872899373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,2,balanced,0.2661706606547038
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,2,balanced,0.16338133811950684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,2,balanced,0.2192373275756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,2,balanced,0.2733813325564067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,2,balanced,0.38049598534901935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,2,balanced,0.48117868105570477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,1,power_law_1.2,52.2803466796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,2,power_law_1.01,3.0005632400512696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,2,balanced,0.7012853622436523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,2,power_law_1.01,0.40945281982421877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,2,balanced,0.9167733192443848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,2,power_law_1.01,0.24780800342559814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,2,balanced,1.3541812896728516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,2,power_law_1.01,0.47432961463928225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,2,power_law_1.01,0.27017600536346437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,2,power_law_1.01,0.2931135892868042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,2,balanced,1.789199988047282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,2,power_law_1.01,0.6122560024261474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,2,power_law_1.01,0.32372479438781737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,2,power_law_1.2,0.19406720399856567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,2,power_law_1.2,0.22695679664611818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,2,power_law_1.01,0.35032958984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,2,power_law_1.2,0.24830079078674316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,2,power_law_1.01,0.6787712097167968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,1,power_law_1.2,16.644927978515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,2,power_law_1.2,0.4075007915496826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,2,power_law_1.2,0.4081088066101074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,2,power_law_1.01,0.6887231826782226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,2,power_law_1.2,0.40993919372558596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,2,power_law_1.2,0.4127935886383057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,2,power_law_1.2,0.42046079635620115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,2,power_law_1.2,0.43506560325622556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,2,power_law_1.2,0.453439998626709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,2,power_law_1.01,0.4327040195465088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,2,power_law_1.2,0.4673664093017578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,2,power_law_1.2,0.4802112102508545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,2,power_law_1.01,0.8729984283447265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,2,power_law_1.01,0.5283391952514649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,2,power_law_1.2,0.48336639404296877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,2,power_law_1.2,0.5726592063903808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,2,power_law_1.01,0.6369919776916504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,2,power_law_1.01,0.051052802801132204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,2,power_law_1.01,1.434291172027588
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,2,power_law_1.01,0.0587007999420166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,2,power_law_1.2,0.6341504096984864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,2,power_law_1.01,0.06986879706382751
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,2,power_law_1.01,0.08579840064048767
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,2,power_law_1.01,0.09034240245819092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,2,power_law_1.2,0.6865664005279541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,2,power_law_1.01,4.922028732299805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,2,power_law_1.01,0.1017408013343811
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,2,power_law_1.01,0.10019840002059936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,2,power_law_1.01,0.10446720123291016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,2,power_law_1.01,0.10712959766387939
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,2,power_law_1.01,0.1103935956954956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,2,power_law_1.01,0.114028799533844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,2,power_law_1.2,0.8295616149902344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,2,power_law_1.01,0.13124480247497558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,2,power_law_1.01,1.7713151931762696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,2,power_law_1.01,0.13731839656829833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,1,power_law_1.2,41.12678527832031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,2,power_law_1.01,0.13121919631958007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,2,power_law_1.01,0.781990385055542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,2,power_law_1.01,0.15639679431915282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,2,power_law_1.2,1.04136323928833
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,2,power_law_1.01,0.18467199802398682
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,2,power_law_1.01,0.17765120267868043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,2,power_law_1.2,1.393113613128662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,1,power_law_1.2,21.181094360351562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,2,power_law_1.01,0.18714879751205443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,2,power_law_1.01,2.248908805847168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,2,power_law_1.01,0.2361407995223999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,2,power_law_1.01,1.1345343589782715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,2,power_law_1.01,1.7682176589965821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,2,power_law_1.2,1.9359872817993165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,2,power_law_1.01,0.32805120944976807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,2,power_law_1.01,3.8814014434814452
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,2,power_law_1.01,0.4033023834228516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,2,power_law_1.01,9.169113922119141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,2,power_law_1.01,2.233011245727539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,2,power_law_1.2,2.405356788635254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,1,power_law_1.2,69.49183959960938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,1,power_law_1.2,30.20279541015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,2,power_law_1.2,0.16833280324935912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,2,power_law_1.2,0.18113919496536254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,2,power_law_1.01,0.5217023849487304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,2,power_law_1.2,0.19244799613952637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,2,power_law_1.2,0.29922559261322024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,2,power_law_1.2,0.3010879993438721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,2,power_law_1.2,0.3067199945449829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,2,power_law_1.2,0.31573119163513186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,2,power_law_1.2,0.3222527980804443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,2,power_law_1.2,0.3392512083053589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,2,power_law_1.2,0.3583296060562134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,2,power_law_1.01,12.561753845214843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,2,power_law_1.2,0.35381760597229006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,2,power_law_1.2,0.3899327993392944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,2,power_law_1.2,0.427020788192749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,2,power_law_1.2,0.48000001907348633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,1,power_law_1.2,54.83036499023437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,2,power_law_1.2,0.5069568157196045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,2,power_law_1.01,0.6402048110961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,2,power_law_1.2,0.6724287986755371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,2,power_law_1.01,4.054751968383789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,2,power_law_1.2,0.13745280504226684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,2,power_law_1.2,0.15603840351104736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,2,power_law_1.2,0.19836800098419188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,2,power_law_1.2,0.7485951900482177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,2,power_law_1.2,0.21338880062103271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,2,power_law_1.2,0.21370880603790282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,2,power_law_1.2,0.2118016004562378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,2,power_law_1.2,0.21458559036254882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,2,power_law_1.2,0.22549118995666503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,2,power_law_1.2,0.2318592071533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,2,power_law_1.01,3.1611007690429687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,2,power_law_1.2,0.2402496099472046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,2,power_law_1.2,0.2504703998565674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,2,power_law_1.2,0.27187199592590333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,2,power_law_1.2,0.2859584093093872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,2,power_law_1.2,0.33712000846862794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,2,power_law_1.2,0.3903615951538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,2,power_law_1.2,3.3305217742919924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,2,power_law_1.2,0.4431488037109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,2,power_law_1.2,0.9634880065917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,2,power_law_1.2,0.5083712100982666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,2,power_law_1.01,4.470028686523437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,2,power_law_1.2,1.1951168060302735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,2,power_law_1.2,0.7371263980865479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,2,power_law_1.01,18.157139587402344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,2,power_law_1.2,1.8809343338012696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,2,power_law_1.2,0.9388480186462402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,2,power_law_1.2,4.827699279785156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,2,power_law_1.01,0.9396927833557129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,2,power_law_1.2,2.2571840286254883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,1,power_law_1.2,39.161746215820315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,2,power_law_1.2,1.1652799606323243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,2,power_law_1.01,6.650777435302734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,2,power_law_1.2,3.018649673461914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,2,power_law_1.2,1.6577152252197265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,2,power_law_1.01,24.334573364257814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,2,1,balanced,0.11773866415023804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,2,1,balanced,0.17721066872278848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,2,1,balanced,0.2756800055503845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,2,1,balanced,0.27959465980529785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,2,1,balanced,0.31622399886449176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,2,1,balanced,0.3158506751060486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,2,1,balanced,0.3189280033111572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,2,1,balanced,0.31775999069213867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,2,1,balanced,0.32215466101964313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,2,1,balanced,0.32344533999760944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,2,1,balanced,0.32658666372299194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,2,1,balanced,0.3330559929211934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,2,1,balanced,0.33768534660339355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,2,1,balanced,0.35276798407236737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,2,1,balanced,0.45189332962036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,2,1,balanced,0.4771413405736287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,2,1,balanced,0.5088853438695272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,2,1,balanced,0.6813759803771973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,2,1,balanced,0.8183146317799886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,2,1,balanced,1.224837303161621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,2,1,balanced,1.5239253044128418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,2,1,balanced,2.3106026649475098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,2,1,balanced,3.1549758911132812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,2,power_law_1.01,6.359558486938477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,2,power_law_1.2,2.3442943572998045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,2,1,balanced,4.898672103881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,2,1,balanced,6.8524214426676435
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,2,power_law_1.2,0.04953599870204926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,2,power_law_1.2,0.05906559824943543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,2,power_law_1.2,0.06420480012893677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,2,power_law_1.2,0.08490880131721497
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,2,power_law_1.2,0.08862720131874084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,2,power_law_1.2,0.10229120254516602
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,2,power_law_1.2,0.10194560289382934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,2,power_law_1.2,0.10549759864807129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,2,power_law_1.2,0.10557440519332886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,2,power_law_1.2,4.710054397583008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,2,power_law_1.2,0.10824960470199585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,2,1,balanced,12.1158078511556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,2,power_law_1.2,0.11276160478591919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,2,power_law_1.01,1.1963135719299316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,2,power_law_1.2,0.13390719890594482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,2,power_law_1.2,0.12332160472869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,2,power_law_1.2,2.841683197021484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,2,power_law_1.2,0.1342463970184326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,2,power_law_1.2,0.14611200094223023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,2,power_law_1.2,0.14332159757614135
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,2,power_law_1.2,0.15623680353164673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,2,power_law_1.2,10.585420989990235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,2,power_law_1.2,0.22512640953063964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,2,1,balanced,17.235178629557293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,2,1,balanced,0.09524266918500264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,2,1,balanced,0.13277332981427512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,2,1,balanced,0.1855306625366211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,2,1,balanced,0.19025067488352457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,2,1,balanced,0.22377600272496542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,2,1,balanced,0.22807466983795166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,2,1,balanced,0.23090134064356485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,2,1,balanced,0.23988266785939535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,2,power_law_1.2,0.2171583890914917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,2,1,balanced,0.24416534105936685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,2,1,balanced,0.24278400341669717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,2,1,balanced,0.2624373237291972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,2,1,balanced,0.2685439984003703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,2,1,balanced,0.2807253400484721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,2,1,balanced,0.3046986659367879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,2,power_law_1.01,8.137958526611328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,2,1,balanced,0.44228267669677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,2,1,balanced,0.4608693520228068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,2,1,balanced,0.47121067841847736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,2,1,balanced,0.6494293212890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,2,1,balanced,0.8105386892954508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,2,1,balanced,1.188383976618449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,2,1,balanced,1.5248053868611653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,2,1,balanced,2.278693358103434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,2,1,balanced,2.9781014124552407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,2,power_law_1.2,0.32433280944824217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,2,1,balanced,4.433338801066081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,2,1,balanced,5.892549514770508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,2,power_law_1.2,5.96099853515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,2,1,balanced,8.804677327473959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,2,power_law_1.2,0.3988991975784302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,2,1,balanced,12.102410634358725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,2,1,balanced,0.08447999755541484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,2,1,balanced,0.11921067039171855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,2,1,balanced,0.17790400981903076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,2,1,balanced,0.17865065733591715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,2,1,balanced,0.18104533354441324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,2,1,balanced,0.18315200010935465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,2,1,balanced,0.18293333053588867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,2,1,balanced,0.18688533703486124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,2,1,balanced,0.1893493334452311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,2,1,balanced,0.1913493275642395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,2,1,balanced,0.198469340801239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,2,1,balanced,0.20359466473261514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,2,1,balanced,0.2104746699333191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,2,1,balanced,0.23092800378799438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,2,1,balanced,0.31545066833496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,2,1,balanced,0.32381866375605267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,2,1,balanced,0.3428373336791992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,2,1,balanced,0.46693865458170575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,2,1,balanced,0.5831093390782675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,2,1,balanced,0.8341653347015381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,2,1,balanced,1.0565919876098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,2,1,balanced,1.5626133282979329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,2,power_law_1.2,4.582720184326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,2,1,balanced,2.0340213775634766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,2,power_law_1.2,13.174234008789062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,2,1,balanced,3.0330079396565757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,2,1,balanced,3.9595947265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,2,power_law_1.2,0.4937856197357178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,2,1,balanced,5.957109451293945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,2,power_law_1.01,1.7221311569213866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,2,1,balanced,7.882277170817058
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,2,power_law_1.2,0.6124671936035156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,2,power_law_1.01,9.245388793945313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,2,1,balanced,0.03923733284076055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,2,1,balanced,0.05923733115196228
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,2,1,balanced,0.08525333801905315
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,2,1,balanced,0.08578667044639587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,2,1,balanced,0.08734933535257976
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,2,1,balanced,0.09038399656613667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,2,1,balanced,0.09902399778366089
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,2,1,balanced,0.10219200452168782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,2,1,balanced,0.10257066289583842
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,2,1,balanced,0.10320533315340678
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,2,1,balanced,0.10539733370145161
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,2,1,balanced,0.11100266377131145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,2,1,balanced,0.11146666606267293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,2,1,balanced,0.11529599626859029
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,2,1,balanced,0.200272003809611
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,2,1,balanced,0.16591466466585794
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,2,1,balanced,0.131221334139506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,2,1,balanced,0.32121066252390545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,2,1,balanced,0.16105066736539206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,2,power_law_1.01,18.09210205078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,2,power_law_1.2,9.273401641845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,2,1,balanced,0.21261866887410483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,2,1,balanced,0.2624053359031677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,2,1,balanced,0.3643306493759155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,2,1,balanced,0.4675306479136149
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,2,1,balanced,0.6786293188730875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,2,1,balanced,0.8950026830037435
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,2,1,balanced,1.322719971338908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,2,1,balanced,1.7497973442077637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,2,1,power_law_1.01,0.10959999561309815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,2,1,power_law_1.01,0.1519871950149536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,2,1,power_law_1.01,0.17280639410018922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,2,1,power_law_1.01,0.26672000885009767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,2,1,power_law_1.01,0.30599679946899416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,2,1,power_law_1.01,0.31258881092071533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,2,1,power_law_1.01,0.3148224115371704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,2,1,power_law_1.01,0.3172352075576782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,2,1,power_law_1.01,0.3281791925430298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,2,1,power_law_1.01,0.3542527914047241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,2,1,power_law_1.01,0.3573632001876831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,2,1,power_law_1.01,0.3747967958450317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,2,1,power_law_1.01,0.38124799728393555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,2,1,power_law_1.01,0.4048319816589355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,2,1,power_law_1.01,0.456441593170166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,2,1,power_law_1.01,0.528934383392334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,2,1,power_law_1.01,0.6081920146942139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,2,1,power_law_1.01,0.7691135883331299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,2,1,power_law_1.01,0.9400639533996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,2,power_law_1.2,17.867884826660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,2,1,power_law_1.01,1.3142848014831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,2,power_law_1.2,5.431564712524414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,2,power_law_1.2,1.0088512420654296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,2,1,power_law_1.01,1.670137596130371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,2,1,power_law_1.01,2.3750335693359377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,2,power_law_1.01,2.4604415893554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,2,1,power_law_1.01,0.09591040015220642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,2,1,power_law_1.01,0.11856000423431397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,2,1,power_law_1.01,0.13583359718322754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,2,1,power_law_1.01,0.1872063994407654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,2,1,power_law_1.01,0.22225279808044435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,2,1,power_law_1.01,0.22544000148773194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,2,1,power_law_1.01,0.23231360912322999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,2,1,power_law_1.01,0.23603200912475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,2,1,power_law_1.01,0.24897921085357666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,2,power_law_1.01,11.337612915039063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,2,1,power_law_1.01,0.27165439128875735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,2,1,power_law_1.01,0.29270401000976565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,2,1,power_law_1.01,0.0828927993774414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,2,1,power_law_1.01,0.11026560068130493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,2,1,power_law_1.01,0.12780159711837769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,2,1,power_law_1.01,0.17622400522232057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,2,1,power_law_1.01,0.1725119948387146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,2,1,power_law_1.01,0.17964160442352295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,2,1,power_law_1.01,0.1807487964630127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,2,1,power_law_1.01,0.17543679475784302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,2,1,power_law_1.01,0.18908159732818602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,2,1,power_law_1.01,0.3184704065322876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,2,1,power_law_1.01,0.19775359630584716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,2,1,power_law_1.01,0.3319936037063599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,2,1,power_law_1.01,0.213919997215271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,2,1,power_law_1.01,3.1068288803100588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,2,1,power_law_1.01,0.3905087947845459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,2,1,power_law_1.01,0.24129281044006348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,2,1,power_law_1.01,0.2515968084335327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,2,1,power_law_1.01,0.4220479965209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,2,1,power_law_1.01,0.27234559059143065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,2,1,power_law_1.01,0.301689600944519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,2,1,power_law_1.01,0.3290816068649292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,2,1,power_law_1.01,0.38127999305725097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,2,power_law_1.2,13.518194580078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,2,power_law_1.01,20.95342102050781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,2,1,power_law_1.01,0.4860991954803467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,2,1,power_law_1.01,0.5528639793395996
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,2,power_law_1.2,1.1751551628112793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,2,1,power_law_1.01,0.7375296115875244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,2,1,power_law_1.01,0.5020287990570068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,2,1,power_law_1.01,0.6165503978729248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,2,1,power_law_1.01,0.9062463760375976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,2,1,power_law_1.01,1.2894463539123535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,2,1,power_law_1.01,0.8605759620666504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,2,1,power_law_1.01,1.1104960441589355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,2,1,power_law_1.01,0.041868799924850465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,2,1,power_law_1.01,0.052985602617263795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,2,1,power_law_1.01,0.06357120275497437
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,2,1,power_law_1.01,0.0844223976135254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,2,1,power_law_1.01,0.08398079872131348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,2,1,power_law_1.01,0.0923520028591156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,2,1,power_law_1.01,0.10169600248336792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,2,1,power_law_1.01,0.10981760025024415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,2,1,power_law_1.01,0.10068479776382447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,2,1,power_law_1.01,0.10443520545959473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,2,1,power_law_1.01,1.5926336288452148
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,2,1,power_law_1.01,0.11431039571762085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,2,1,power_law_1.01,1.6498624801635742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,2,1,power_law_1.01,0.1096127986907959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,2,1,power_law_1.01,0.11367039680480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,2,1,power_law_1.01,0.12791680097579955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,2,1,power_law_1.01,0.12817920446395875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,2,1,power_law_1.01,0.13433599472045898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,2,1,power_law_1.01,0.14803199768066405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,2,1,power_law_1.01,0.20208001136779785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,2,1,power_law_1.01,4.928428649902344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,2,1,power_law_1.01,0.19540480375289918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,2,1,power_law_1.01,2.088915252685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,2,1,power_law_1.01,2.3692352294921877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,2,1,power_law_1.01,0.2873471975326538
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,2,1,power_law_1.01,0.29901440143585206
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,2,1,power_law_1.01,0.4050879955291748
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,2,1,power_law_1.01,0.5256127834320068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,2,1,power_law_1.01,3.0926591873168947
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,2,1,power_law_1.01,0.7238336086273194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,2,1,power_law_1.01,0.924403190612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,2,1,power_law_1.01,3.058323287963867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,2,1,power_law_1.01,4.0489662170410154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,2,power_law_1.2,19.477381896972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,2,1,power_law_1.2,0.1102336049079895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,2,1,power_law_1.2,0.1382464051246643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,2,1,power_law_1.2,0.1876736044883728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,2,1,power_law_1.2,0.25764479637146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,2,1,power_law_1.2,0.31101438999176023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,2,1,power_law_1.2,0.3137727975845337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,2,1,power_law_1.2,0.31512320041656494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,2,1,power_law_1.2,0.31674239635467527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,2,1,power_law_1.2,0.3289088010787964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,2,1,power_law_1.2,0.35882880687713625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,2,1,power_law_1.2,0.3687488079071045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,2,1,power_law_1.2,0.37421441078186035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,2,1,power_law_1.2,0.37852160930633544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,2,1,power_law_1.2,0.41866240501403806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,2,1,power_law_1.2,0.47090559005737304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,2,1,power_law_1.2,0.555020809173584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,2,1,power_law_1.2,0.6085824012756348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,2,power_law_1.2,10.2080322265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,2,1,power_law_1.01,4.554150390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,2,1,power_law_1.2,0.772819185256958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,2,1,power_law_1.2,0.9561535835266113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,2,1,power_law_1.01,6.653766632080078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,2,power_law_1.2,1.6391872406005858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,2,1,power_law_1.2,1.3064512252807616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,2,power_law_1.2,31.53665771484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,2,1,power_law_1.2,0.0948736011981964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,2,1,power_law_1.2,0.12181119918823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,2,1,power_law_1.2,0.14584959745407106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,2,1,power_law_1.2,0.18661760091781615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,2,1,power_law_1.2,0.22221438884735106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,2,1,power_law_1.2,0.22566399574279786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,2,1,power_law_1.2,0.23071999549865724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,2,1,power_law_1.2,0.2369663953781128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,2,1,power_law_1.2,0.24222080707550048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,2,1,power_law_1.2,0.27308800220489504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,2,1,power_law_1.2,0.2886399984359741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,2,1,power_law_1.2,0.32194559574127196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,2,1,power_law_1.2,0.33466238975524903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,2,1,power_law_1.2,0.38577919006347655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,2,1,power_law_1.2,0.42439680099487304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,2,1,power_law_1.2,1.6551231384277343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,2,1,power_law_1.2,0.491871976852417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,2,1,power_law_1.01,1.371116828918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,2,1,power_law_1.01,6.006316757202148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,2,1,power_law_1.2,0.5512447834014893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,2,1,power_law_1.01,6.036038589477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,2,1,power_law_1.2,0.7522111892700195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,2,1,power_law_1.2,0.9096639633178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,2,1,power_law_1.2,1.273977565765381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,2,1,power_law_1.2,1.6492095947265626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,2,1,power_law_1.2,2.404204750061035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,2,1,power_law_1.01,7.972659301757813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,2,1,power_law_1.2,0.08284800052642823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,2,1,power_law_1.2,0.1097599983215332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,2,1,power_law_1.2,0.11491199731826782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,2,1,power_law_1.2,0.1717311978340149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,2,1,power_law_1.2,0.17299840450286866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,2,1,power_law_1.2,0.17912960052490234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,2,1,power_law_1.2,0.18029439449310303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,2,1,power_law_1.2,0.18309760093688965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,2,1,power_law_1.2,0.18919039964675904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,2,1,power_law_1.2,2.3766912460327148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,2,1,power_law_1.2,0.20194559097290038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,2,1,power_law_1.2,0.21258881092071533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,2,1,power_law_1.2,0.23825280666351317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,2,1,power_law_1.2,0.24600958824157715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,2,1,power_law_1.2,0.2712575912475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,2,1,power_law_1.2,0.30024960041046145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,2,1,power_law_1.2,0.3276927947998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,2,1,power_law_1.2,0.383404803276062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,2,1,power_law_1.01,8.936434936523437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,2,1,power_law_1.2,0.49397759437561034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,2,1,power_law_1.2,0.6127744197845459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,2,1,power_law_1.01,11.214342498779297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,2,power_law_1.2,11.822457885742187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,2,1,power_law_1.2,0.8583999633789062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,2,1,power_law_1.01,1.8843711853027343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,2,balanced,0.21526400248209634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,2,balanced,0.32815466324488324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,2,balanced,0.5004160006841024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,2,balanced,0.6375893354415894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,2,balanced,0.637989322344462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,2,balanced,0.6296746730804443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,2,balanced,0.64246932665507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,2,balanced,0.6463573376337687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,2,balanced,0.6218986511230469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,2,balanced,0.6566613515218099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,2,balanced,0.6230026483535767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,2,balanced,0.6191360155741373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,2,1,power_law_1.2,1.1148223876953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,2,balanced,0.6274400154749552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,2,balanced,0.655626654624939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,2,balanced,0.783141295115153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,2,balanced,0.7977386315663656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,2,balanced,0.8248106638590494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,2,balanced,1.3352853457132976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,2,balanced,1.5625227292378743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,2,1,power_law_1.2,3.1078208923339843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,2,balanced,2.2840906778971353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,2,1,power_law_1.2,3.134681510925293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,2,balanced,2.924912134806315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,2,1,power_law_1.01,11.89336929321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,2,balanced,0.18604799111684164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,2,balanced,0.23411200443903604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,2,balanced,0.3280533353487651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,2,balanced,0.42046932379404706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,2,balanced,0.4195679823557536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,2,balanced,4.704127947489421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,2,balanced,0.4117813507715861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,2,balanced,0.41699198881785077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,2,balanced,0.42762664953867596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,2,1,power_law_1.2,0.04062080085277557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,2,balanced,0.4405226707458496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,2,1,power_law_1.2,0.05082880258560181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,2,balanced,0.437552014986674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,2,1,power_law_1.2,0.06155520081520081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,2,balanced,0.4477866490681966
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,2,1,power_law_1.2,0.08380159735679626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,2,1,power_law_1.2,0.08487039804458618
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,2,1,power_law_1.2,0.09191039800643921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,2,balanced,6.569301605224609
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,2,1,power_law_1.2,0.10353920459747315
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,2,1,power_law_1.2,0.11018879413604736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,2,1,power_law_1.2,0.09973120093345642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,2,1,power_law_1.2,4.574995040893555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,2,1,power_law_1.2,0.10583679676055908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,2,1,power_law_1.2,0.1161471962928772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,2,balanced,0.47601600488026935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,2,balanced,0.5074346860249838
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,2,1,power_law_1.2,0.11171200275421142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,2,balanced,0.5506453514099121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,2,balanced,0.7038079897562662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,2,1,power_law_1.2,0.11579519510269165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,2,balanced,0.7356320222218832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,2,balanced,0.792634646097819
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,2,1,power_law_1.2,0.13187839984893798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,2,1,power_law_1.2,1.5997695922851562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,2,balanced,11.450421651204428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,2,balanced,1.1665386358896892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,2,1,power_law_1.2,0.1267583966255188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,2,balanced,1.4107573827107747
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,2,1,power_law_1.2,0.14202239513397216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,2,balanced,2.0725653966267905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,2,balanced,2.6844746271769204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,2,balanced,4.030506769816081
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,2,1,power_law_1.2,0.15192960500717162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,2,balanced,15.917157491048178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,2,balanced,5.293317476908366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,2,1,power_law_1.2,2.086131286621094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,2,1,power_law_1.2,0.16284799575805664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,2,balanced,8.000282923380533
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,2,1,power_law_1.2,0.19572479724884034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,2,balanced,11.87997817993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,2,balanced,25.35003153483073
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,2,1,power_law_1.2,0.2504447937011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,2,1,power_law_1.01,15.601951599121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,2,balanced,0.1551253298918406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,2,balanced,0.21067200104395548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,2,balanced,0.29997867345809937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,2,balanced,0.30188266436258954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,2,balanced,0.30058133602142334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,2,1,power_law_1.2,0.3094208002090454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,2,balanced,19.58135477701823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,2,balanced,0.30499200026194256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,2,balanced,0.30807467301686603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,2,balanced,0.31251200040181476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,2,balanced,0.32042133808135986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,2,balanced,0.3261813322703044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,2,balanced,0.3312693238258362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,2,balanced,0.3472586472829183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,2,balanced,0.35439467430114746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,2,balanced,0.39475735028584796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,2,balanced,0.5434186855951945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,2,power_law_1.2,2.417024040222168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,2,balanced,0.5647146701812744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,2,balanced,0.5918986797332764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,2,balanced,35.00316874186198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,2,balanced,0.763487974802653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,2,power_law_1.01,0.303219199180603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,2,balanced,0.9824373722076416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,2,power_law_1.01,0.30723841190338136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,2,power_law_1.01,0.3551552057266235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,2,balanced,1.3948319753011067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,2,power_law_1.01,0.6354688167572021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,2,balanced,1.864202658335368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,2,power_law_1.01,0.6351488113403321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,2,1,power_law_1.2,4.877900695800781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,2,power_law_1.01,0.6372479915618896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,2,balanced,2.708245277404785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,2,power_law_1.01,0.6437952041625976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,2,power_law_1.01,0.6475008010864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,2,1,power_law_1.2,6.041785430908203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,2,power_law_1.01,0.6535808086395264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,2,balanced,3.5570398966471353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,2,power_law_1.01,0.6573887825012207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,2,power_law_1.01,0.6556928157806396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,2,power_law_1.01,0.6892159938812256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,2,1,power_law_1.2,0.39943039417266846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,2,balanced,5.289829254150391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,2,power_law_1.01,0.7260479927062988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,2,balanced,29.83550516764323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,2,power_law_1.01,0.2021951913833618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,2,power_law_1.01,0.7665535926818847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,2,power_law_1.01,0.24289920330047607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,2,power_law_1.01,0.24316799640655518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,2,power_law_1.01,1.089254379272461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,2,balanced,6.956576029459636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,2,power_law_1.01,1.127359962463379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,2,power_law_1.01,1.315980815887451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,2,power_law_1.01,1.5871295928955078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,2,power_law_1.01,0.42133121490478515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,2,power_law_1.01,0.4185919761657715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,2,power_law_1.01,0.4150847911834717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,2,balanced,10.472325642903646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,2,power_law_1.01,0.4275519847869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,2,power_law_1.01,0.454252815246582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,2,1,power_law_1.2,3.0630399703979494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,2,power_law_1.01,0.466431999206543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,2,power_law_1.01,2.207891273498535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,2,power_law_1.01,0.4761343955993652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,2,power_law_1.01,0.5447872161865235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,2,power_law_1.01,0.5627776145935058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,2,power_law_1.01,0.6310719966888427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,2,power_law_1.01,0.7211455821990966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,2,power_law_1.01,0.902560043334961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,2,balanced,13.875204722086588
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,2,1,power_law_1.2,0.5214015960693359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,2,power_law_1.01,0.9915328025817871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,2,power_law_1.01,3.080940818786621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,2,power_law_1.01,0.16522879600524903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,2,power_law_1.01,1.0761664390563965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,2,power_law_1.01,0.22990720272064208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,2,power_law_1.01,0.2596927881240845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,2,power_law_1.01,0.2978816032409668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,2,power_law_1.01,0.29894399642944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,2,power_law_1.01,0.3027647972106934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,2,power_law_1.01,1.7196735382080077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,2,power_law_1.01,0.306329607963562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,2,power_law_1.01,0.3125375986099243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,2,power_law_1.01,0.318886399269104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,2,power_law_1.01,0.3385215997695923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,2,power_law_1.01,4.224012756347657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,2,power_law_1.01,0.35808000564575193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,2,power_law_1.01,0.4082943916320801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,2,power_law_1.01,1.9253311157226562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,2,power_law_1.01,0.45014400482177735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,2,power_law_1.01,0.5370175838470459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,2,power_law_1.01,0.5891007900238037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,2,power_law_1.01,0.6481919765472413
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,2,balanced,0.05579199890295664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,2,power_law_1.01,0.7410111904144288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,2,balanced,0.07956266899903615
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,2,balanced,0.12680000066757202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,2,balanced,0.126720001300176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,2,power_law_1.01,2.371494483947754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,2,power_law_1.01,1.0602239608764648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,2,balanced,0.12917866309483847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,2,balanced,0.1304639975229899
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,2,balanced,0.13154666622479758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,2,balanced,0.1323306659857432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,2,balanced,0.13802133003870645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,2,balanced,0.137882669766744
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,2,balanced,0.14044800400733948
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,2,balanced,0.16793066263198853
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,2,balanced,0.1597760021686554
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,2,balanced,0.15848533312479654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,2,balanced,0.3376693328221639
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,2,balanced,0.2449493408203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,2,balanced,0.16546133160591125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,2,balanced,0.36538668473561603
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,2,balanced,0.25333333015441895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,2,balanced,0.34143467744191486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,2,power_law_1.01,3.812006378173828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,2,balanced,0.4546346664428711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,2,balanced,0.6431039969126383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,2,balanced,0.8555946350097656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,2,balanced,1.2789706389109294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,2,balanced,1.713007926940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,2,power_law_1.01,1.3320192337036132
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,2,balanced,2.5883359909057617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,2,balanced,3.465786616007487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,2,power_law_1.01,6.66798095703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,2,power_law_1.01,1.7451648712158203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,2,power_law_1.01,5.6383617401123045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,2,1,power_law_1.2,6.71783676147461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,2,power_law_1.01,2.286854362487793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,2,1,power_law_1.2,0.7303552150726318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,2,power_law_1.01,9.346438598632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,2,1,power_law_1.2,4.042579269409179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,2,power_law_1.01,6.549267578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,2,power_law_1.01,3.4796672821044923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,2,power_law_1.01,0.0626688003540039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,2,power_law_1.01,0.08256639838218689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,2,power_law_1.01,10.680480194091796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,2,power_law_1.01,4.532870483398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,2,1,power_law_1.2,0.9215935707092285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,2,power_law_1.01,0.10975359678268433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,2,power_law_1.01,0.12597119808197021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,2,power_law_1.01,0.13218560218811035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,2,1,power_law_1.2,8.970265960693359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,2,power_law_1.01,0.17031680345535277
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,2,power_law_1.01,0.15634560585021973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,2,power_law_1.01,0.1793727993965149
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,2,power_law_1.01,0.16521600484848023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,2,power_law_1.01,0.16461440324783325
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,2,power_law_1.01,0.18916480541229247
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,2,power_law_1.01,0.1800447940826416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,2,power_law_1.01,0.19708160161972046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,2,power_law_1.01,0.23244800567626953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,2,power_law_1.01,0.19911680221557618
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,2,power_law_1.01,0.2325439929962158
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,2,power_law_1.01,0.23320960998535156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,2,power_law_1.01,7.168505859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,2,power_law_1.01,0.326528000831604
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,2,power_law_1.01,0.3821631908416748
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,2,power_law_1.01,0.5217984199523926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,2,power_law_1.01,0.5755712032318115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,2,power_law_1.01,17.225375366210937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,2,power_law_1.01,14.86339874267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,2,power_law_1.01,0.9096256256103515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,2,1,power_law_1.2,11.032128143310548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,2,power_law_1.01,8.595193481445312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,2,power_law_1.01,1.1381759643554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,2,power_law_1.01,19.98395538330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,2,power_law_1.01,29.984429931640626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,2,power_law_1.01,14.225132751464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,2,power_law_1.01,1.854374313354492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,2,power_law_1.01,2.299711990356445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,2,1,power_law_1.2,1.364851188659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,2,1,power_law_1.2,15.73516845703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,2,power_law_1.2,0.2546303987503052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,2,power_law_1.2,0.28257920742034914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,2,power_law_1.2,0.3493184089660645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,2,power_law_1.2,0.636678409576416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,2,1,power_law_1.2,5.999942398071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,2,power_law_1.2,0.6406784057617188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,2,power_law_1.2,0.6387519836425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,2,power_law_1.2,0.6429120063781738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,2,power_law_1.2,0.648473596572876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,2,power_law_1.2,0.6523200035095215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,2,power_law_1.2,0.6539904117584229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,2,power_law_1.2,0.6601920127868652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,2,power_law_1.2,0.7080704212188721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,2,power_law_1.2,0.7208384037017822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,2,power_law_1.2,0.7978559970855713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,2,power_law_1.2,0.9123583793640136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,2,power_law_1.01,3.8037441253662108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,2,power_law_1.2,1.044057559967041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,2,power_law_1.2,1.3208319664001464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,2,1,power_law_1.2,12.018732452392578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,2,power_law_1.2,0.20185599327087403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,2,power_law_1.2,0.21283199787139892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,2,power_law_1.2,0.2716671943664551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,2,power_law_1.2,0.4088895797729492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,2,power_law_1.2,0.4176959991455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,2,power_law_1.2,0.41921281814575195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,2,power_law_1.2,1.787654495239258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,2,power_law_1.2,0.4401535987854004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,2,power_law_1.2,0.45072641372680666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,2,power_law_1.2,0.471999979019165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,2,power_law_1.2,0.5187456130981445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,2,power_law_1.2,2.090015983581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,2,power_law_1.2,0.5229951858520507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,2,power_law_1.2,0.5631231784820556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,2,power_law_1.2,0.6087679862976074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,2,power_law_1.2,0.7490047931671142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,2,power_law_1.2,0.8279168128967285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,2,power_law_1.2,3.1198144912719727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,2,power_law_1.2,1.0336383819580077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,2,power_law_1.2,1.161734390258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,2,power_law_1.2,3.7729278564453126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,2,power_law_1.2,1.4346431732177733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,2,power_law_1.01,33.86394348144531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,2,power_law_1.2,0.17469439506530762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,2,power_law_1.2,0.23086719512939452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,2,power_law_1.2,0.23378560543060303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,2,power_law_1.2,0.29719679355621337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,2,power_law_1.2,0.2986112117767334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,2,power_law_1.2,0.3023871898651123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,2,power_law_1.2,0.3066751956939697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,2,power_law_1.2,0.31103360652923584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,2,power_law_1.2,0.33523199558258054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,2,power_law_1.2,0.3436543941497803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,2,power_law_1.2,2.0925119400024412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,2,power_law_1.2,0.3532799959182739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,2,power_law_1.2,0.4299327850341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,2,power_law_1.2,0.4365248203277588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,2,power_law_1.2,0.5162879943847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,2,power_law_1.2,5.928070449829102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,2,power_law_1.2,0.5661119937896728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,2,power_law_1.2,0.6057087898254394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,2,power_law_1.01,38.11991577148437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,2,power_law_1.2,0.7487808227539062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,2,power_law_1.2,2.9311424255371095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,2,power_law_1.2,1.0436991691589355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,2,power_law_1.2,1.2884991645812989
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,2,1,power_law_1.2,1.843289566040039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,2,power_law_1.2,4.029177474975586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,2,power_law_1.2,1.9851520538330079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,2,power_law_1.2,8.57791976928711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,2,power_law_1.2,2.456012725830078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,2,power_law_1.2,5.188742446899414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,2,power_law_1.2,3.5243904113769533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,2,power_law_1.2,7.510246276855469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,2,power_law_1.2,0.06898559927940369
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,2,power_law_1.2,0.0757311999797821
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,2,power_law_1.2,0.10288000106811523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,2,power_law_1.2,0.12748160362243652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,2,power_law_1.2,0.12914559841156006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,2,power_law_1.2,0.17080960273742676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,2,power_law_1.2,4.492454528808594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,2,power_law_1.2,0.1700543999671936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,2,power_law_1.01,4.523660659790039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,2,1,balanced,0.16861865917841592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,2,1,balanced,0.2841920057932536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,2,1,balanced,0.5526560147603353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,2,1,balanced,0.4843200047810872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,2,1,balanced,0.4978346824645996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,2,1,balanced,0.5013173421223959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,2,1,balanced,0.540890653928121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,2,1,balanced,0.5372746785481771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,2,1,balanced,0.5348480145136515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,2,1,balanced,0.4914826552073161
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,2,power_law_1.2,0.1855296015739441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,2,1,balanced,0.49899200598398846
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,2,power_law_1.2,0.15998719930648803
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,2,power_law_1.2,0.1678976058959961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,2,power_law_1.2,0.19503999948501588
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,2,power_law_1.2,0.19761279821395875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,2,power_law_1.2,0.18970240354537965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,2,power_law_1.2,17.767832946777343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,2,power_law_1.2,0.2166975975036621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,2,1,balanced,0.5065226554870605
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,2,power_law_1.2,0.2773504018783569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,2,1,balanced,0.5153813362121582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,2,1,balanced,0.5276533365249634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,2,1,balanced,0.7663946946461996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,2,power_law_1.01,17.224595642089845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,2,power_law_1.2,0.20115840435028076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,2,1,balanced,0.7773066361745199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,2,1,balanced,0.12301333745320638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,2,1,balanced,0.18361065785090128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,2,1,balanced,0.8129013379414877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,2,1,balanced,0.3117706576983134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,2,1,balanced,0.31141867240269977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,2,1,balanced,1.1389866669972737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,2,1,balanced,1.383402665456136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,2,power_law_1.2,0.3241472005844116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,2,1,power_law_1.2,7.963667297363282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,2,1,balanced,2.0381226539611816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,2,1,balanced,0.10742933551470439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,2,1,balanced,0.16113066673278809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,2,1,balanced,0.2934719920158386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,2,1,balanced,0.2956800063451131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,2,1,balanced,0.2965493400891622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,2,1,balanced,0.29895999034245807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,2,1,balanced,0.30137600501378375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,2,1,balanced,0.30424533287684125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,2,1,balanced,0.30931733051935834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,2,1,balanced,0.31249066193898517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,2,1,balanced,0.3190133372942607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,2,1,balanced,0.3257066607475281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,2,1,balanced,0.3314080039660136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,2,1,balanced,0.35972265402475995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,2,1,balanced,0.46371734142303467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,2,1,balanced,2.8359785079956055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,2,1,balanced,0.47951467831929523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,2,power_law_1.2,0.3207040071487427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,2,1,balanced,0.4991573492685954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,2,1,balanced,0.6898667017618815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,2,1,balanced,0.8864320119222006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,2,1,balanced,3.820224126180013
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,2,power_law_1.2,0.3557760000228882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,2,1,balanced,1.2776106993357341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,2,1,balanced,1.6503359476725261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,2,1,balanced,0.31360000371932983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,2,1,balanced,2.43230930964152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,2,1,balanced,5.600613276163737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,2,1,balanced,3.17576535542806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,2,1,balanced,0.31940267483393353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,2,1,balanced,4.729429244995117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,2,1,balanced,8.51746686299642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,2,1,balanced,0.3259359995524089
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,2,power_law_1.2,0.5422592163085938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,2,1,balanced,0.335749348004659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,2,1,balanced,0.3485013246536255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,2,1,balanced,0.3495200077692668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,2,1,balanced,0.3643946647644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,2,1,balanced,0.3853493531545003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,2,1,balanced,0.3986826737721761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,2,1,balanced,0.43826134999593097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,2,1,balanced,0.6504319906234741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,2,1,balanced,6.193301518758138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,2,1,balanced,0.6675626436869303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,2,1,balanced,0.7244799931844076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,2,1,balanced,1.0269493261973064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,2,1,balanced,1.3077600002288818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,2,1,balanced,12.698591868082682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,2,1,balanced,9.304858525594076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,2,1,balanced,1.9345547358194988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,2,power_law_1.2,0.6521279811859131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,2,1,balanced,2.495680014292399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,2,1,balanced,3.727802594502767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,2,1,balanced,12.38797378540039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,2,power_law_1.2,6.448255920410157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,2,1,balanced,20.902154286702473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,2,1,balanced,4.900618553161621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,2,power_law_1.2,11.183283233642578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,2,1,balanced,7.298351923624675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,2,power_law_1.01,46.098104858398436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,2,1,power_law_1.01,0.15212160348892212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,2,1,power_law_1.01,0.25654399394989014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,2,1,power_law_1.01,0.3042304039001465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,2,1,power_law_1.01,0.44063358306884765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,2,1,power_law_1.01,0.5234752178192139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,2,power_law_1.2,19.646194458007812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,2,1,power_law_1.01,0.5048255920410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,2,1,power_law_1.01,0.543507194519043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,2,1,balanced,9.69549306233724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,2,1,power_law_1.01,0.5401599884033204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,2,1,power_law_1.01,0.5499711990356445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,2,1,power_law_1.01,0.5544960021972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,2,1,power_law_1.01,0.5830976009368897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,2,1,power_law_1.01,0.6563392162322998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,2,power_law_1.2,0.8041088104248046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,2,1,power_law_1.01,0.6799615859985352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,2,1,power_law_1.01,0.748576021194458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,2,1,power_law_1.01,0.7668352127075195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,2,1,balanced,32.22910817464193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,2,1,power_law_1.01,0.12086399793624877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,2,1,power_law_1.01,0.8662079811096192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,2,1,balanced,14.50707753499349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,2,1,power_law_1.01,0.1743551969528198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,2,1,power_law_1.01,0.22522239685058593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,2,1,power_law_1.01,0.29524478912353513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,2,1,power_law_1.01,0.31337599754333495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,2,1,power_law_1.01,1.0186944007873535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,2,1,power_law_1.01,0.3230079889297485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,2,1,balanced,0.05400000015894572
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,2,1,balanced,0.08121066788832347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,2,1,power_law_1.01,0.3374079942703247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,2,1,power_law_1.01,0.3659008026123047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,2,1,power_law_1.01,0.38280320167541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,2,1,power_law_1.01,0.4008639812469482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,2,1,power_law_1.01,0.460262393951416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,2,1,power_law_1.01,1.2834752082824707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,2,1,power_law_1.01,0.495084810256958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,2,1,power_law_1.01,0.5371647834777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,2,1,balanced,0.12583466370900473
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,2,1,balanced,0.12468799948692322
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,2,1,balanced,0.12652800480524698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,2,1,power_law_1.01,0.576473617553711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,2,1,balanced,0.12889066338539124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,2,1,balanced,0.14526399970054626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,2,1,balanced,0.14968533317248026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,2,1,power_law_1.01,1.6166656494140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,2,1,balanced,0.14965867002805075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,2,1,balanced,0.1511360009511312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,2,1,balanced,0.15260799725850424
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,2,1,balanced,0.15999466180801392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,2,1,balanced,0.1596160034338633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,2,1,balanced,0.16688533624013266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,2,1,balanced,0.26052266359329224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,2,1,balanced,0.21731199820836386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,2,1,balanced,0.18756266434987387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,2,1,balanced,0.4124373197555542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,2,1,balanced,0.23170665899912515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,2,1,balanced,22.39881642659505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,2,1,balanced,0.3140373428662618
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,2,1,balanced,0.3973919947942098
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,2,1,balanced,0.566373348236084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,2,1,power_law_1.01,0.6404672145843506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,2,1,power_law_1.01,2.200998306274414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,2,1,power_law_1.01,0.10367360115051269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,2,1,power_law_1.01,0.15164799690246583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,2,1,power_law_1.01,0.7517183780670166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,2,1,power_law_1.01,0.1827839970588684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,2,1,power_law_1.01,0.2848959922790527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,2,1,power_law_1.01,0.8683839797973633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,2,1,power_law_1.01,0.28999040126800535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,2,1,power_law_1.01,0.2964224100112915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,2,1,power_law_1.01,1.1698431968688965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,2,1,power_law_1.01,0.29592959880828856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,2,1,power_law_1.01,0.3014976024627686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,2,1,power_law_1.01,0.30273919105529784
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,2,1,balanced,0.7364853223164877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,2,1,power_law_1.01,0.3071295976638794
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,2,1,balanced,1.0786720116933186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,2,1,power_law_1.01,0.3193984031677246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,2,1,power_law_1.01,0.3390399932861328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,2,1,balanced,1.4258240063985188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,2,power_law_1.2,8.607923126220703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,2,1,power_law_1.01,1.4465791702270507
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,2,1,balanced,2.109498659769694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,2,1,power_law_1.01,2.8263296127319335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,2,1,balanced,2.850192070007324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,2,1,power_law_1.01,0.3501823902130127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,2,1,power_law_1.01,0.3983360052108765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,2,1,power_law_1.01,0.46090240478515626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,2,1,power_law_1.01,0.48889598846435545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,2,1,power_law_1.01,0.567136001586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,2,1,power_law_1.01,0.75731201171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,2,1,power_law_1.01,2.0478656768798826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,2,1,power_law_1.01,4.13834228515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,2,1,power_law_1.01,0.9447872161865234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,2,1,power_law_1.01,1.3285759925842284
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,2,power_law_1.2,1.1804224014282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,2,1,power_law_1.01,2.6579967498779298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,2,power_law_1.2,15.084556579589844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,2,1,power_law_1.01,5.731795120239258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,2,1,power_law_1.01,0.05189120173454285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,2,1,power_law_1.01,0.07008640170097351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,2,1,power_law_1.01,1.6970048904418946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,2,1,power_law_1.01,3.8605121612548827
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,2,1,power_law_1.01,0.085971200466156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,2,1,power_law_1.01,0.11986559629440308
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,2,1,power_law_1.01,0.1266111969947815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,2,1,power_law_1.01,0.13701759576797484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,2,1,power_law_1.01,0.14886399507522582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,2,1,power_law_1.01,0.1700543999671936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,2,1,power_law_1.01,0.14885760545730592
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,2,1,power_law_1.01,0.15589760541915892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,2,1,power_law_1.01,0.1769279956817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,2,1,power_law_1.01,0.16099200248718262
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,2,1,power_law_1.01,0.17638399600982665
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,2,1,power_law_1.01,0.19738880395889283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,2,1,power_law_1.01,0.1894271969795227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,2,1,power_law_1.01,2.4766208648681642
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,2,1,power_law_1.01,0.20500481128692627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,2,1,power_law_1.01,0.2362623929977417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,2,1,power_law_1.01,5.0531455993652346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,2,power_law_1.2,1.8930496215820312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,2,1,power_law_1.01,0.2599744081497192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,2,power_law_1.2,14.136441040039063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,2,1,power_law_1.01,8.765331268310547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,2,1,power_law_1.01,0.3290047883987427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,2,1,power_law_1.01,0.38562560081481934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,2,1,power_law_1.01,7.478054046630859
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,2,1,power_law_1.01,0.4586304187774658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,2,1,power_law_1.01,3.2342208862304687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,2,1,power_law_1.01,0.6311488151550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,2,1,power_law_1.01,0.803872013092041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,2,1,power_law_1.01,9.84988784790039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,2,1,power_law_1.01,1.1584704399108887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,2,power_law_1.2,36.320562744140624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,2,1,power_law_1.01,12.450259399414062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,2,1,power_law_1.01,4.752633666992187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,2,1,power_law_1.01,1.4590784072875977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,2,power_law_1.2,2.211180877685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,2,1,power_law_1.01,14.63573760986328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,2,1,power_law_1.01,6.2792510986328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,2,power_law_1.2,30.355975341796874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,2,1,power_law_1.01,2.1380224227905273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,2,1,power_law_1.01,9.30466537475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,2,1,power_law_1.01,20.30005798339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,2,1,power_law_1.01,20.032563781738283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,2,1,power_law_1.2,0.14974720478057862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,2,1,power_law_1.2,0.22110719680786134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,2,1,power_law_1.2,0.30636799335479736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,2,1,power_law_1.2,0.45424637794494627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,2,1,power_law_1.2,0.5102591991424561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,2,1,power_law_1.2,0.5116928100585938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,2,1,power_law_1.2,0.5417344093322753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,2,1,power_law_1.2,0.5439743995666504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,2,power_law_1.2,48.260659790039064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,2,1,power_law_1.2,0.5491519927978515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,2,1,power_law_1.2,0.12163200378417968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,2,1,power_law_1.2,0.546950387954712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,2,power_law_1.2,20.536344909667967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,2,1,power_law_1.2,0.16890239715576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,2,1,power_law_1.2,0.5624703884124755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,2,1,power_law_1.2,0.19567359685897828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,2,1,power_law_1.2,0.30909440517425535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,2,1,power_law_1.2,0.3116544008255005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,2,1,power_law_1.2,0.10466560125350952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,2,1,power_law_1.2,0.319488000869751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,2,1,power_law_1.2,0.14994560480117797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,2,1,power_law_1.2,0.34179201126098635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,2,1,power_law_1.2,0.18913279771804808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,2,1,power_law_1.2,0.35997440814971926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,2,1,power_law_1.2,0.2923392057418823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,2,1,power_law_1.2,0.39855360984802246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,2,1,power_law_1.2,0.28793599605560305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,2,1,power_law_1.2,0.29297919273376466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,2,1,power_law_1.2,0.29589760303497314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,2,1,power_law_1.2,0.29828479290008547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,2,1,power_law_1.2,0.30373759269714357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,2,1,power_law_1.2,0.635916805267334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,2,1,power_law_1.2,0.3179903984069824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,2,1,power_law_1.2,0.6800896167755127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,2,1,power_law_1.2,0.31687040328979493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,2,1,power_law_1.2,0.33948800563812254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,2,1,power_law_1.2,0.3496320009231567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,2,1,power_law_1.2,0.4192319869995117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,2,1,power_law_1.2,0.4576128005981445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,2,1,power_law_1.2,0.744870376586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,2,1,power_law_1.2,0.43982720375061035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,2,1,power_law_1.2,0.44736638069152834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,2,1,power_law_1.2,0.7845248222351074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,2,1,power_law_1.2,0.504582405090332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,2,1,power_law_1.2,0.8547648429870606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,2,1,power_law_1.2,0.526700782775879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,2,1,power_law_1.2,0.5806528091430664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,2,power_law_1.2,3.319968032836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,2,1,power_law_1.2,0.9943103790283203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,2,1,power_law_1.2,0.4845952033996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,2,1,power_law_1.2,0.6343552112579346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,2,1,power_law_1.2,1.2666879653930665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,2,1,power_law_1.2,0.7039936065673829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,2,1,power_law_1.2,0.5784255981445312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,2,1,power_law_1.2,0.8619520187377929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,2,1,power_law_1.2,1.162713623046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,2,1,power_law_1.2,0.7497087955474854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,2,1,power_law_1.2,1.448844814300537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,2,1,power_law_1.2,1.5798975944519043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,2,1,power_law_1.01,2.912473678588867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,2,1,power_law_1.2,0.9287232398986817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,2,1,power_law_1.01,26.883807373046874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,2,1,power_law_1.2,2.0288639068603516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,2,1,power_law_1.2,2.1535104751586913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,2,balanced,0.11483200391133626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,2,balanced,0.16331199804941812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,2,1,power_law_1.2,1.3445695877075194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,2,balanced,0.3045813242594401
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,2,balanced,0.5308213233947754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,2,balanced,0.9589707056681315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,2,balanced,1.8337546984354656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,2,balanced,1.8425973256429036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,2,balanced,1.8455413182576497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,2,balanced,1.8497546513875325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,2,1,power_law_1.2,2.7705408096313477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,2,balanced,1.8820640246073406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,2,balanced,1.8696266810099285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,2,1,power_law_1.2,2.648236846923828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,2,1,power_law_1.2,1.7083648681640624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,2,balanced,2.1062026023864746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,2,1,power_law_1.01,12.346803283691406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,2,1,power_law_1.2,0.05242239832878113
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,2,1,power_law_1.2,0.0702015995979309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,2,balanced,0.09319999814033508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,2,1,power_law_1.2,0.08794879913330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,2,balanced,0.13168533643086752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,2,balanced,0.19909866650899252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,2,1,power_law_1.2,0.12132480144500732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,2,balanced,0.330677330493927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,2,1,power_law_1.2,0.1225600004196167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,2,balanced,0.6138666470845541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,2,1,power_law_1.2,0.1377087950706482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,2,1,power_law_1.2,0.14926719665527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,2,1,power_law_1.2,0.16974079608917236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,2,balanced,2.1131253242492676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,2,balanced,2.1359200477600098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,2,balanced,2.159872055053711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,2,1,power_law_1.2,2.477132797241211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,2,balanced,1.1829333305358887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,2,balanced,2.178592046101888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,2,balanced,1.1894880135854085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,2,1,power_law_1.2,0.14758399724960328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,2,balanced,1.1919093132019043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,2,1,power_law_1.2,0.16151039600372313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,2,balanced,2.2216533025105796
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,2,1,power_law_1.2,0.17760640382766724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,2,balanced,1.1958133379618328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,2,balanced,1.2018986543019612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,2,1,power_law_1.2,0.16506240367889405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,2,balanced,1.2101279894510906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,2,1,power_law_1.2,4.2357440948486325
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,2,1,power_law_1.2,0.17582080364227295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,2,balanced,1.32695468266805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,2,balanced,1.3435306549072266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,2,1,power_law_1.2,0.20049281120300294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,2,balanced,1.366773287455241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,2,balanced,1.397648016611735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,2,1,power_law_1.2,0.19435520172119142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,2,balanced,2.2973705927530923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,2,1,power_law_1.2,0.2015615940093994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,2,balanced,1.4253120422363281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,2,1,power_law_1.2,0.23310720920562744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,2,balanced,1.4738186200459797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,2,balanced,1.5991147359212239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,2,balanced,1.6994880040486653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,2,1,power_law_1.2,0.24198400974273682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,2,balanced,1.9273974100748699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,2,balanced,2.3473013242085776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,2,1,power_law_1.2,3.8251392364501955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,2,balanced,2.167861302693685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,2,1,power_law_1.2,0.31766400337219236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,2,balanced,2.5147627194722495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,2,balanced,3.34661865234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,2,balanced,2.720992088317871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,2,balanced,3.8531786600748696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,2,balanced,3.9051198959350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,2,balanced,5.584746678670247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,2,balanced,4.451279958089192
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,2,1,power_law_1.2,0.4044288158416748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,2,1,power_law_1.2,3.2209793090820313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,2,balanced,7.27238400777181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,2,balanced,6.451679865519206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,2,1,power_law_1.2,5.04598388671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,2,1,power_law_1.2,5.509088134765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,2,balanced,8.464096069335938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,2,balanced,10.707412719726562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,2,1,power_law_1.2,0.4534207820892334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,2,balanced,14.683226267496744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,2,balanced,14.464154561360678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,2,balanced,0.08762133121490479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,2,balanced,0.11216533184051514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,2,balanced,0.18545067310333252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,2,balanced,0.3304906686147054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,2,balanced,0.5970666805903116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,2,balanced,1.1486026446024578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,2,1,power_law_1.2,0.6260799884796142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,2,balanced,1.157322645187378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,2,balanced,1.1616480350494385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,2,balanced,1.1680746873219807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,2,balanced,1.176042636235555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,2,1,power_law_1.2,4.776665496826172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,2,balanced,21.931294759114582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,2,balanced,1.1878666877746582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,2,balanced,1.1997493108113606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,2,balanced,1.2118186950683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,2,balanced,1.2338026364644368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,2,balanced,1.2576053142547607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,2,balanced,1.2781600157419841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,2,balanced,1.3211039702097576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,2,balanced,1.3981119791666667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,2,balanced,1.4711626370747883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,2,balanced,1.618245283762614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,2,balanced,1.8604639371236165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,2,balanced,2.671354611714681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,2,balanced,3.047178586324056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,2,balanced,0.03982399900754293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,2,balanced,0.05765333275000254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,2,balanced,0.08794666330019633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,2,balanced,0.1435413360595703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,2,balanced,0.2540480097134908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,2,balanced,0.46399466196695965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,2,balanced,0.4707839886347453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,2,balanced,0.4757973353068034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,2,balanced,0.47758932908376056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,2,balanced,0.48201600710550946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,2,balanced,0.48818135261535645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,2,1,power_law_1.2,0.7945663928985596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,2,balanced,0.4944586753845215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,2,balanced,4.3029225667317705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,2,balanced,0.5049493312835693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,2,balanced,0.5174986521402994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,2,balanced,0.5080000162124634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,2,balanced,0.5163840055465698
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,2,balanced,0.5352640151977539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,2,balanced,0.5454879999160767
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,2,balanced,0.5682933330535889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,2,balanced,5.56985600789388
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,2,balanced,0.5941439867019653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,2,power_law_1.2,5.624940872192383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,2,balanced,0.6264640092849731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,2,1,power_law_1.2,7.451667022705078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,2,power_law_1.2,44.723489379882814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,2,balanced,0.7863732973734537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,2,power_law_1.01,0.17415039539337157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,2,power_law_1.01,0.31312639713287355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,2,power_law_1.01,0.4196352005004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,2,balanced,8.158512115478516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,2,power_law_1.01,0.6256256103515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,2,power_law_1.01,0.8423040390014649
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,2,balanced,0.7435466448465983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,2,1,power_law_1.2,1.1455103874206543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,2,balanced,1.3247893651326497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,2,1,power_law_1.2,8.725299072265624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,2,balanced,1.225530703862508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,2,power_law_1.01,0.11547520160675048
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,2,balanced,1.7431519826253254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,2,power_law_1.01,0.1977728009223938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,2,balanced,10.690347035725912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,2,power_law_1.01,0.26698238849639894
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,2,balanced,2.269877274831136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,2,power_law_1.01,0.4362495899200439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,2,power_law_1.01,1.0602368354797362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,2,power_law_1.01,0.5483391761779786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,2,power_law_1.01,1.4278079986572265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,2,power_law_1.01,0.7276800155639649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,2,power_law_1.01,1.4349056243896485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,2,power_law_1.01,1.5622143745422363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,2,power_law_1.01,0.10747519731521607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,2,power_law_1.01,0.18273279666900635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,2,power_law_1.01,1.4969087600708009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,2,power_law_1.01,0.25948801040649416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,2,power_law_1.01,0.3991424083709717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,2,power_law_1.01,0.9452735900878906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,2,power_law_1.01,1.5988287925720215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,2,power_law_1.01,0.547488021850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,2,power_law_1.01,0.9712575912475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,2,power_law_1.01,0.6762879848480224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,2,power_law_1.01,1.861369514465332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,2,power_law_1.01,1.020198440551758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,2,power_law_1.01,0.9341376304626465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,2,power_law_1.01,1.0059840202331543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,2,power_law_1.01,1.7903936386108399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,2,power_law_1.01,0.9566847801208496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,2,power_law_1.01,1.101676845550537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,2,power_law_1.01,0.9391231536865234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,2,power_law_1.01,1.909529685974121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,2,power_law_1.01,1.1311296463012694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,2,power_law_1.01,1.026963233947754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,2,power_law_1.01,1.9442623138427735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,2,power_law_1.01,1.1493760108947755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,2,power_law_1.01,0.05681920051574707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,2,power_law_1.01,0.9679295539855957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,2,1,power_law_1.2,1.4912320137023927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,2,power_law_1.01,1.2667903900146484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,2,power_law_1.01,1.0114944458007813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,2,power_law_1.01,1.95665283203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,2,power_law_1.01,1.3370752334594727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,2,power_law_1.01,1.073190402984619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,2,power_law_1.01,2.07891845703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,2,power_law_1.01,1.3182463645935059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,2,power_law_1.01,1.4328255653381348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,2,power_law_1.01,2.2495807647705077
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,2,power_law_1.01,0.08302080035209655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,2,power_law_1.01,0.11204479932785034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,2,power_law_1.01,1.7017663955688476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,2,power_law_1.01,0.17935359477996826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,2,power_law_1.01,2.506771278381348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,2,power_law_1.01,0.2248447895050049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,2,power_law_1.01,1.1084608078002929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,2,power_law_1.01,1.8878656387329102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,2,power_law_1.01,0.27498879432678225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,2,power_law_1.01,1.1597567558288575
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,2,power_law_1.01,0.36785919666290284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,2,power_law_1.01,2.9406015396118166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,2,power_law_1.01,0.398528003692627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,2,power_law_1.01,2.3279296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,2,power_law_1.01,1.1885760307312012
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,2,power_law_1.01,0.39661440849304197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,2,power_law_1.01,1.2638079643249511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,2,power_law_1.01,3.3524097442626952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,2,power_law_1.01,2.7847360610961913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,2,power_law_1.01,1.3910207748413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,2,power_law_1.01,1.555776023864746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,2,power_law_1.01,3.611577606201172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,2,power_law_1.01,4.423052978515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,2,power_law_1.01,0.398086404800415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,2,power_law_1.01,1.8823104858398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,2,power_law_1.01,0.41866240501403806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,2,power_law_1.01,0.4378687858581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,2,power_law_1.01,2.214259147644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,2,power_law_1.01,0.45802879333496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,2,power_law_1.01,4.629753494262696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,2,power_law_1.01,0.49483518600463866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,2,power_law_1.01,0.46460800170898436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,2,power_law_1.01,2.802617645263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,2,power_law_1.01,0.49214720726013184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,2,1,power_law_1.2,6.293580627441406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,2,power_law_1.01,0.5472064018249512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,2,power_law_1.01,0.5435264110565186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,2,power_law_1.01,6.406163024902344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,2,power_law_1.01,3.554617691040039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,2,power_law_1.01,5.448179244995117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,2,power_law_1.01,0.6220223903656006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,2,power_law_1.01,0.645907211303711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,2,power_law_1.01,0.7462143898010254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,2,power_law_1.01,4.963091278076172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,2,power_law_1.01,7.306041717529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,2,power_law_1.01,8.392704010009766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,2,power_law_1.01,0.8742400169372558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,2,power_law_1.01,6.242464065551758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,2,power_law_1.01,0.963475227355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,2,power_law_1.01,1.2573951721191405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,2,power_law_1.01,9.241343688964843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,2,power_law_1.01,1.508556842803955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,2,1,power_law_1.2,9.8216064453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,2,power_law_1.01,11.971040344238281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,2,power_law_1.01,8.922962951660157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,2,power_law_1.01,2.1322303771972657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,2,power_law_1.01,14.035308837890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,2,power_law_1.01,3.579654312133789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,2,power_law_1.01,11.632883453369141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,2,power_law_1.2,0.1523967981338501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,2,power_law_1.2,0.30138239860534666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,2,power_law_1.2,0.11716480255126953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,2,power_law_1.01,15.56420440673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,2,power_law_1.2,0.3829888105392456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,2,power_law_1.2,0.19861119985580444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,2,power_law_1.2,0.5399231910705566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,2,power_law_1.2,0.2460479974746704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,2,power_law_1.2,0.8002431869506836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,2,power_law_1.2,0.9059007644653321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,2,power_law_1.2,1.3579903602600099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,2,power_law_1.2,1.4526144027709962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,2,power_law_1.2,0.3954240083694458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,2,power_law_1.2,0.5290239810943603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,2,power_law_1.2,1.4303680419921876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,2,power_law_1.01,18.011045837402342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,2,power_law_1.2,0.6543231964111328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,2,power_law_1.2,1.5612288475036622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,2,power_law_1.2,0.9067647933959961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,2,1,power_law_1.2,12.465650939941407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,2,power_law_1.2,0.983456039428711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,2,power_law_1.2,1.6125759124755858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,2,power_law_1.2,0.9833024024963379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,2,power_law_1.2,1.8134464263916015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,2,power_law_1.2,0.944979190826416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,2,power_law_1.2,1.8047872543334962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,2,power_law_1.2,1.1093952178955078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,2,power_law_1.2,1.1588735580444336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,2,power_law_1.2,1.9420799255371093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,2,power_law_1.2,1.22607364654541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,2,power_law_1.2,1.9878719329833985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,2,power_law_1.2,1.3202239990234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,2,power_law_1.2,2.0709056854248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,2,power_law_1.2,1.3190655708312988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,2,power_law_1.2,1.3818431854248048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,2,power_law_1.2,2.122380828857422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,2,power_law_1.2,1.5358592033386231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,2,power_law_1.2,2.3534719467163088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,2,power_law_1.2,1.7436864852905274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,2,power_law_1.2,0.057036799192428586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,2,power_law_1.2,2.571327972412109
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,2,power_law_1.2,0.08554880023002624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,2,power_law_1.2,0.1030079960823059
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,2,power_law_1.2,0.16475520133972169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,2,power_law_1.2,3.021753692626953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,2,power_law_1.2,0.22243199348449708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,2,power_law_1.2,0.26634879112243653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,2,power_law_1.2,1.9150144577026367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,2,power_law_1.2,0.37144320011138915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,2,power_law_1.2,3.4820415496826174
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,2,power_law_1.2,0.38052480220794677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,2,power_law_1.2,0.3762176036834717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,2,power_law_1.2,2.4458688735961913
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,2,power_law_1.2,0.40853118896484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,2,power_law_1.2,0.40819201469421384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,2,power_law_1.2,4.433311843872071
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,2,power_law_1.2,0.4452928066253662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,2,power_law_1.2,2.841638374328613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,2,power_law_1.2,0.44371838569641114
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,2,power_law_1.2,0.5096447944641114
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,2,power_law_1.2,0.48526721000671386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,2,power_law_1.2,3.820230484008789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,2,power_law_1.2,0.5051136016845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,2,power_law_1.2,5.952755355834961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,2,power_law_1.2,0.5522240161895752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,2,power_law_1.2,0.5788544178009033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,2,power_law_1.2,0.6625472068786621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,2,power_law_1.2,4.7985279083251955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,2,1,power_law_1.2,9.350867462158202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,2,power_law_1.2,0.6792255878448487
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,2,1,power_law_1.2,2.162995147705078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,2,power_law_1.2,0.7963520050048828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,2,power_law_1.2,7.500428771972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,2,power_law_1.2,0.9183872222900391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,2,power_law_1.2,1.106214427947998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,2,power_law_1.2,6.599327850341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,2,power_law_1.2,1.3665408134460448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,2,power_law_1.2,1.698828887939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,2,power_law_1.2,9.87852783203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,2,power_law_1.2,2.2579456329345704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,2,power_law_1.2,8.550962829589844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,2,power_law_1.2,0.10714240074157715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,2,power_law_1.2,0.1818560004234314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,2,power_law_1.2,3.8862079620361327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,2,1,balanced,0.09909333785374959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,2,1,power_law_1.2,19.56621398925781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,2,power_law_1.2,0.2512383937835693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,2,power_law_1.2,14.372909545898438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,2,power_law_1.2,0.34530560970306395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,2,1,balanced,0.1420693298180898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,2,1,balanced,0.2856266697247823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,2,power_law_1.2,0.49074559211730956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,2,1,balanced,0.5077173312505087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,2,power_law_1.2,13.099494934082031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,2,1,balanced,0.9515306949615479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,2,power_law_1.2,0.7012671947479248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,2,1,balanced,1.8502613703409831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,2,power_law_1.2,0.905907154083252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,2,1,balanced,1.85535462697347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,2,power_law_1.2,0.9056127548217774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,2,1,balanced,1.8567040761311848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,2,power_law_1.2,0.9636608123779297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,2,1,balanced,1.8585920333862305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,2,power_law_1.2,0.9880703926086426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,2,1,balanced,1.8703306516011555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,2,power_law_1.2,1.0180992126464843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,2,1,balanced,1.8740746180216472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,2,power_law_1.2,1.048684787750244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,2,1,power_law_1.2,14.637158203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,2,1,balanced,1.8822827339172363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,2,power_law_1.2,1.074944019317627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,2,1,balanced,1.9178773562113445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,2,power_law_1.2,1.1160256385803222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,2,power_law_1.2,15.682791137695313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,2,power_law_1.2,1.1710016250610351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,2,1,balanced,0.0864586631457011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,2,1,balanced,0.10938666264216106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,2,1,balanced,0.19471999009450278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,2,1,balanced,0.33056533336639404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,2,power_law_1.2,1.1946880340576171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,2,1,balanced,0.6275360186894735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,2,1,balanced,1.2094826698303223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,2,power_law_1.2,1.2963775634765624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,2,power_law_1.2,18.567916870117188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,2,1,balanced,1.2129066785176594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,2,1,balanced,1.2167200247446697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,2,power_law_1.2,1.4721280097961427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,2,1,balanced,1.2186346848805745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,2,1,balanced,1.2216906547546387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,2,power_law_1.2,1.5972415924072265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,2,1,balanced,1.228058656056722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,2,1,balanced,1.236506700515747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,2,1,balanced,1.9147520065307617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,2,power_law_1.2,1.9361600875854492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,2,1,balanced,1.2431893348693848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,2,1,balanced,2.079589366912842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,2,1,balanced,1.254426638285319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,2,1,balanced,0.07828266421953838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,2,1,balanced,0.10565867026646932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,2,1,balanced,1.3201119899749756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,2,1,balanced,2.093503952026367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,2,1,balanced,1.3431466420491536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,2,1,balanced,2.126666704813639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,2,1,balanced,1.3839306831359863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,2,1,balanced,1.496869405110677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,2,1,balanced,2.1985546747843423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,2,1,balanced,1.5708692868550618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,2,1,balanced,2.2628159523010254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,2,power_law_1.2,2.261612892150879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,2,1,balanced,1.767306645711263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,2,1,balanced,2.416330655415853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,2,1,balanced,1.9709760348002117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,2,power_law_1.2,2.995590400695801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,2,1,balanced,2.6486239433288574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,2,1,balanced,3.0519946416219077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,2,1,balanced,0.18148799737294516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,2,1,balanced,0.3323199947675069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,2,1,balanced,0.6121546824773153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,2,1,balanced,3.7867466608683267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,2,1,balanced,3.4948641459147134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,2,power_law_1.2,3.6665599822998045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,2,1,balanced,1.1760106881459553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,2,1,balanced,1.1827413241068523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,2,1,balanced,1.1874079704284668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,2,1,balanced,4.0971574783325195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,2,1,balanced,5.0264692306518555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,2,1,balanced,1.1931733290354412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,2,1,balanced,1.1977813243865967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,2,power_law_1.2,5.077292633056641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,2,1,balanced,1.2087999979654949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,2,1,balanced,1.2184373537699382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,2,1,balanced,6.069333394368489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,2,1,balanced,1.2296000321706135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,2,1,balanced,6.548559824625651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,2,1,balanced,1.2465919653574626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,2,1,balanced,1.2626667022705078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,2,1,balanced,1.2809866269429524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,2,power_law_1.2,6.6829887390136715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,2,1,balanced,1.313530683517456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,2,1,balanced,7.837210973103841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,2,1,balanced,1.3761280377705891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,2,1,balanced,1.4290720621744792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,2,1,balanced,9.640560150146484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,2,1,balanced,1.546709378560384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,2,1,balanced,1.7730186780293782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,2,1,balanced,2.593290646870931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,2,power_law_1.2,9.497222137451171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,2,1,balanced,11.736335754394531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,2,1,balanced,2.890138626098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,2,1,balanced,12.72769037882487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,2,1,balanced,4.107040087381999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,2,1,balanced,5.286837259928386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,2,power_law_1.2,12.008710479736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,2,1,balanced,17.96083704630534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,2,1,power_law_1.01,0.09745280146598816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,2,1,balanced,7.714901606241862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,2,1,power_law_1.01,0.14082560539245606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,2,1,power_law_1.01,0.08584960103034973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,2,1,power_law_1.01,0.2963007926940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,2,1,power_law_1.01,0.11075839996337891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,2,1,power_law_1.01,0.5245823860168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,2,1,power_law_1.01,0.19307520389556884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,2,1,power_law_1.01,0.7470079898834229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,2,1,power_law_1.01,0.3290047883987427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,2,1,power_law_1.01,0.48578557968139646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,2,1,power_law_1.01,0.9858112335205078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,2,1,power_law_1.01,0.6374783992767334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,2,1,balanced,10.12063980102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,2,1,power_law_1.01,1.360217571258545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,2,1,power_law_1.01,0.8895039558410645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,2,1,power_law_1.01,1.4223103523254395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,2,1,power_law_1.01,0.9311424255371094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,2,1,power_law_1.01,0.96810884475708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,2,1,power_law_1.01,1.4871040344238282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,2,1,balanced,0.03937600056330363
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,2,1,balanced,0.058506667613983154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,2,1,power_law_1.01,0.9797311782836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,2,1,power_law_1.01,1.4935744285583497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,2,1,power_law_1.01,1.0430208206176759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,2,1,power_law_1.01,1.5538240432739259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,2,1,power_law_1.01,1.1156479835510253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,2,1,power_law_1.01,1.648806381225586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,2,1,power_law_1.01,1.1774080276489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,2,1,balanced,0.08627733588218689
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,2,1,balanced,0.14081066846847534
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,2,1,balanced,0.24877333641052246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,2,1,power_law_1.01,1.3385087966918945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,2,1,power_law_1.01,1.7012031555175782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,2,1,power_law_1.01,1.2302399635314942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,2,1,power_law_1.01,1.8475711822509766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,2,1,power_law_1.01,1.2623680114746094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,2,1,power_law_1.01,1.853811264038086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,2,1,power_law_1.01,1.3560447692871094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,2,1,balanced,0.4638559818267822
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,2,1,balanced,0.47069334983825684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,2,1,power_law_1.01,1.8534400939941407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,2,1,balanced,0.4758400122324626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,2,1,balanced,0.4790560007095337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,2,1,power_law_1.01,0.08533759713172913
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,2,1,balanced,0.48205868403116864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,2,1,balanced,0.48813867568969727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,2,1,power_law_1.01,1.9936895370483398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,2,1,balanced,0.49427199363708496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,2,1,balanced,0.5016586780548096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,2,1,balanced,0.514794667561849
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,2,1,balanced,0.5502719879150391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,2,1,power_law_1.01,2.121977615356445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,2,1,balanced,0.5613119999567667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,2,1,balanced,0.5861333211263021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,2,1,power_law_1.01,1.5332608222961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,2,1,balanced,0.591375986735026
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,2,1,balanced,0.6229066848754883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,2,1,power_law_1.01,2.321356773376465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,2,1,balanced,0.6858773231506348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,2,1,power_law_1.01,0.1047104001045227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,2,1,power_law_1.01,1.7236928939819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,2,1,balanced,0.7343786557515463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,2,1,power_law_1.01,0.18029439449310303
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,2,1,balanced,0.9055573145548502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,2,1,power_law_1.01,0.32990078926086425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,2,1,balanced,0.9370720386505127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,2,1,power_law_1.01,2.6848960876464845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,2,1,power_law_1.01,2.0869823455810548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,2,1,power_law_1.01,0.4879039764404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,2,1,balanced,1.732602596282959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,2,1,power_law_1.01,0.6649407863616943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,2,1,balanced,1.3047093550364177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,2,1,power_law_1.01,2.4566976547241213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,2,1,power_law_1.01,3.1219072341918945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,2,1,power_law_1.01,0.8875519752502441
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,2,1,balanced,1.8498239517211914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,2,1,power_law_1.01,0.9353856086730957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,2,1,balanced,2.4030027389526367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,2,1,power_law_1.01,3.2331966400146483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,2,1,power_law_1.2,20.826329040527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,2,1,power_law_1.01,0.9735936164855957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,2,1,power_law_1.01,3.9917057037353514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,2,1,power_law_1.2,0.09875199794769288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,2,1,power_law_1.01,0.9861120223999024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,2,1,power_law_1.2,0.14099839925765992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,2,1,power_law_1.01,1.009727954864502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,2,1,power_law_1.2,0.2827455997467041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,2,1,power_law_1.01,3.9882686614990233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,2,1,power_law_1.2,0.4905087947845459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,2,1,power_law_1.01,1.04335355758667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,2,1,power_law_1.2,28.02471008300781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,2,1,power_law_1.2,0.6425536155700684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,2,1,power_law_1.01,4.7058368682861325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,2,1,power_law_1.01,1.0775872230529786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,2,1,power_law_1.2,0.08597760200500489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,2,1,power_law_1.01,1.1080639839172364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,2,1,power_law_1.2,0.10937600135803223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,2,1,power_law_1.01,5.533983993530273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,2,1,power_law_1.2,0.1947648048400879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,2,1,power_law_1.01,1.1293567657470702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,2,1,power_law_1.2,0.3300607919692993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,2,1,power_law_1.01,1.171622371673584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,2,1,power_law_1.2,0.4436863899230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,2,1,power_law_1.01,6.4676673889160154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,2,1,power_law_1.2,0.8914943695068359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,2,1,power_law_1.2,0.5881152153015137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,2,1,power_law_1.01,1.2307711601257325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,2,1,power_law_1.2,0.8967807769775391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,2,1,power_law_1.2,1.3342720031738282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,2,1,power_law_1.01,1.369702434539795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,2,1,power_law_1.01,7.08485107421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,2,1,power_law_1.2,0.9188032150268555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,2,1,power_law_1.2,1.3919424057006835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,2,1,power_law_1.01,1.5218303680419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,2,1,power_law_1.2,0.972544002532959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,2,1,power_law_1.01,0.03597440123558045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,2,1,power_law_1.2,1.4445247650146484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,2,1,power_law_1.2,0.9924863815307617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,2,1,power_law_1.01,1.7954879760742188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,2,1,power_law_1.01,8.420172882080077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,2,1,power_law_1.2,1.499392032623291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,2,1,power_law_1.2,1.076531219482422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,2,1,power_law_1.01,2.0676223754882814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,2,1,power_law_1.2,1.1492735862731933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,2,1,power_law_1.2,1.534995174407959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,2,1,power_law_1.2,1.2148608207702636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,2,1,power_law_1.01,10.16663055419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,2,1,power_law_1.01,2.677177619934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,2,1,power_law_1.01,0.05751039981842041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,2,1,power_law_1.2,1.3899007797241212
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,2,1,power_law_1.01,0.08503040075302123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,2,1,power_law_1.01,0.13861119747161865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,2,1,power_law_1.2,1.253657627105713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,2,1,power_law_1.01,0.20120959281921386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,2,1,power_law_1.01,3.2785598754882814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,2,1,power_law_1.2,1.277785587310791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,2,1,power_law_1.01,0.2671231985092163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,2,1,power_law_1.2,1.6252735137939454
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,2,1,power_law_1.01,0.3564224004745483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,2,1,power_law_1.2,1.4175871849060058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,2,1,power_law_1.01,12.022335815429688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,2,1,power_law_1.01,0.3801919937133789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,2,1,power_law_1.2,1.7158655166625976
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,2,1,power_law_1.01,0.38747520446777345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,2,1,power_law_1.01,4.484844970703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,2,1,power_law_1.01,0.4014912128448486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,2,1,power_law_1.2,1.9071744918823241
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,2,1,power_law_1.01,0.41155200004577636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,2,1,power_law_1.01,0.42046079635620115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,2,1,power_law_1.01,13.250802612304687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,2,1,power_law_1.2,1.8863296508789062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,2,1,power_law_1.2,1.5866175651550294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,2,1,power_law_1.01,5.670905685424804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,2,1,power_law_1.2,1.9628543853759766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,2,1,power_law_1.2,1.7980928421020508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,2,1,power_law_1.2,2.0042560577392576
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,2,1,power_law_1.01,0.43528318405151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,2,1,power_law_1.2,2.158835220336914
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,2,1,power_law_1.01,0.4734015941619873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,2,1,power_law_1.2,2.2160831451416017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,2,1,power_law_1.01,0.5100287914276123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,2,1,power_law_1.2,0.0768064022064209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,2,1,power_law_1.01,15.659429931640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,2,1,power_law_1.01,0.5499839782714844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,2,1,power_law_1.2,0.1048192024230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,2,1,power_law_1.2,2.533363151550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,2,1,power_law_1.01,8.07110366821289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,2,1,power_law_1.2,2.4495487213134766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,2,1,power_law_1.2,0.17937920093536378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,2,1,power_law_1.01,0.6393919944763183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,2,1,power_law_1.2,2.965376091003418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,2,1,power_law_1.2,0.3191551923751831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,2,1,power_law_1.01,0.6025599956512451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,2,balanced,0.09727999567985535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,2,balanced,0.13918399810791016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,2,balanced,0.26709334055582684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,2,1,power_law_1.2,0.44158082008361815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,2,balanced,0.4626346826553345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,2,1,power_law_1.2,3.3036991119384767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,2,1,power_law_1.2,2.7592063903808595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,2,balanced,0.8201279640197754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,2,1,power_law_1.2,0.612281608581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,2,balanced,1.5744907061258953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,2,1,power_law_1.2,0.8678208351135254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,2,balanced,1.5773013432820637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,2,1,power_law_1.2,3.1400575637817383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,2,balanced,1.5826560656229656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,2,1,power_law_1.01,0.7012864112854004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,2,balanced,1.5879573822021484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,2,1,power_law_1.01,0.7163072109222413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,2,balanced,1.5868959426879883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,2,1,power_law_1.2,4.05096321105957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,2,1,power_law_1.01,0.848089599609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,2,balanced,1.5980000495910645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,2,1,power_law_1.01,0.9060735702514648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,2,balanced,1.8128426869710286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,2,1,power_law_1.2,4.083212661743164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,2,balanced,1.8171839714050293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,2,balanced,1.8348479270935059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,2,1,power_law_1.2,5.094393539428711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,2,balanced,1.8491360346476238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,2,balanced,1.867311954498291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,2,1,power_law_1.2,0.9209535598754883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,2,1,power_law_1.2,0.037171199917793274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,2,1,power_law_1.01,1.0637120246887206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,2,balanced,1.9010559717814128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,2,1,power_law_1.2,0.9536000251770019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,2,1,power_law_1.01,10.429625701904296
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,2,1,power_law_1.01,1.2430848121643066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,2,balanced,1.9804587364196777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,2,1,power_law_1.2,6.688838195800781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,2,balanced,0.08348799745241801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,2,balanced,0.11920533577601115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,2,balanced,2.034224033355713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,2,balanced,0.17933867375055948
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,2,1,power_law_1.01,1.5133184432983398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,2,balanced,0.29133333762486774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,2,1,power_law_1.2,5.639360046386718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,2,balanced,0.5320746501286825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,2,balanced,2.1855626106262207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,2,1,power_law_1.2,12.392537689208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,2,balanced,1.0228746732076008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,2,1,power_law_1.2,0.056620800495147706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,2,balanced,1.0281439622243245
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,2,1,power_law_1.01,2.0511295318603517
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,2,1,power_law_1.2,0.08408960103988647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,2,balanced,2.377456029256185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,2,balanced,1.030682643254598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,2,1,power_law_1.2,0.9551936149597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,2,1,power_law_1.2,0.1244928002357483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,2,balanced,1.033450682957967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,2,1,power_law_1.2,1.0030847549438477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,2,balanced,3.4563093185424805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,2,1,power_law_1.01,3.044748878479004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,2,1,power_law_1.2,1.0544639587402345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,2,1,power_law_1.2,7.199180603027344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,2,1,power_law_1.2,1.0737919807434082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,2,balanced,4.181893348693848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,2,balanced,0.08313600222269694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,2,1,power_law_1.2,1.106060791015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,2,balanced,0.10420266787211101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,2,balanced,0.16423466801643372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,2,balanced,0.28938666979471844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,2,1,power_law_1.2,0.18399360179901122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,2,balanced,0.5166079998016357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,2,1,power_law_1.2,1.155129623413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,2,1,power_law_1.2,8.48370590209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,2,balanced,0.993285338083903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,2,balanced,1.0006133715311687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,2,balanced,5.797791798909505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,2,balanced,1.0048747062683105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,2,balanced,1.012885332107544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,2,balanced,1.0170880158742268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,2,balanced,1.039621353149414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,2,balanced,1.029088020324707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,2,balanced,1.0485119819641113
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,2,1,power_law_1.2,0.24531199932098388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,2,balanced,1.1473546822865803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,2,balanced,7.596410751342773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,2,1,power_law_1.2,1.1840959548950196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,2,1,power_law_1.2,0.35359361171722414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,2,balanced,1.158138672510783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,2,1,power_law_1.2,0.3640448093414307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,2,balanced,1.1815253098805745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,2,1,power_law_1.2,1.2798527717590331
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,2,1,power_law_1.2,0.39336318969726564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,2,balanced,1.2166720231374104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,2,1,power_law_1.2,10.324185943603515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,2,1,power_law_1.2,0.3990911960601807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,2,balanced,1.2340373198191326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,2,1,power_law_1.2,12.144659423828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,2,balanced,1.0414986610412598
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,2,balanced,0.035562666753927864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,2,1,power_law_1.2,0.409606409072876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,2,balanced,0.054431999723116554
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,2,balanced,0.08054399987061818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,2,balanced,1.051594654719035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,2,balanced,0.127210666735967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,2,1,power_law_1.2,0.4162303924560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,2,balanced,1.0726453463236492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,2,1,power_law_1.2,0.44159998893737795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,2,balanced,1.0928479830423992
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,2,1,power_law_1.2,0.482528018951416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,2,balanced,1.1114346981048584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,2,balanced,12.12829844156901
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,2,1,power_law_1.2,0.5348288059234619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,2,balanced,1.1521226565043132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,2,1,power_law_1.2,1.4199935913085937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,2,balanced,1.2222399711608887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,2,balanced,1.2919306755065918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,2,balanced,1.29148268699646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,2,balanced,1.391541322072347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,2,balanced,1.495914618174235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,2,balanced,1.6963946024576824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,2,1,power_law_1.2,13.447782897949219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,2,1,power_law_1.2,0.5672383785247803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,2,balanced,1.9066880544026692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,2,1,power_law_1.2,1.5811903953552247
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,2,1,power_law_1.2,0.6825215816497803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,2,power_law_1.01,0.1282495975494385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,2,1,power_law_1.2,15.801280212402343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,2,1,power_law_1.2,0.6450175762176513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,2,balanced,1.422826608022054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,2,1,power_law_1.2,1.8518016815185547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,2,1,power_law_1.2,0.7351295948028564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,2,power_law_1.01,0.11150720119476318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,2,balanced,1.5988000233968098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,2,balanced,16.691477457682293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,2,power_law_1.01,0.18439680337905884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,2,1,power_law_1.2,0.753446388244629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,2,1,power_law_1.2,2.158176040649414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,2,balanced,0.2241226633389791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,2,1,power_law_1.2,0.8841728210449219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,2,balanced,0.4034186601638794
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,2,balanced,0.40753066539764404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,2,balanced,2.9339466094970703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,2,power_law_1.01,0.2546880006790161
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,2,balanced,0.41225067774454754
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,2,1,power_law_1.2,0.9338432312011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,2,1,power_law_1.2,2.746905517578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,2,1,power_law_1.2,1.1101887702941895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,2,balanced,3.3933547337849936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,2,balanced,2.3598079681396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,2,power_law_1.01,0.0996288001537323
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,2,1,power_law_1.2,1.2881855964660645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,2,1,power_law_1.2,3.3381183624267576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,2,power_law_1.01,0.15809279680252075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,2,balanced,2.6544906298319497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,2,power_law_1.01,0.22854399681091309
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,2,balanced,0.4146346648534139
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,2,1,power_law_1.2,1.5571200370788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,2,power_law_1.01,0.35662720203399656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,2,power_law_1.01,0.3570175886154175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,2,power_law_1.01,0.5567039966583252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,2,power_law_1.01,0.476416015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,2,power_law_1.01,0.2509567975997925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,2,power_law_1.01,0.7437376022338867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,2,1,power_law_1.2,2.1104511260986327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,2,power_law_1.01,0.3694911956787109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,2,power_law_1.01,0.6242815971374511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,2,power_law_1.01,0.8647744178771972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,2,balanced,4.899973233540853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,2,power_law_1.01,0.4982463836669922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,2,balanced,0.41716798146565753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,2,power_law_1.01,0.8026495933532715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,2,balanced,0.42342400550842285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,2,power_law_1.01,1.194662380218506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,2,power_law_1.01,0.5348608016967773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,2,balanced,0.43084267775217694
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,2,balanced,0.4369279940923055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,2,balanced,3.819242795308431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,2,balanced,0.45114131768544513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,2,power_law_1.01,0.7964159965515136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,2,power_law_1.01,1.2703295707702638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,2,1,power_law_1.2,4.498310470581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,2,power_law_1.01,0.8731200218200683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,2,balanced,6.386175791422526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,2,power_law_1.01,1.3632448196411133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,2,power_law_1.01,0.8088128089904785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,2,balanced,4.935488065083821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,2,power_law_1.01,1.293836784362793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,2,power_law_1.01,0.8584383964538574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,2,1,power_law_1.2,3.204006576538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,2,power_law_1.01,1.3802047729492188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,2,power_law_1.01,0.8982975959777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,2,1,power_law_1.2,5.704902267456054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,2,power_law_1.01,0.9868288040161133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,2,power_law_1.01,1.5612031936645507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,2,balanced,7.2868906656901045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,2,balanced,9.391141255696615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,2,power_law_1.01,1.0527680397033692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,2,power_law_1.01,1.5657535552978517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,2,power_law_1.01,1.0434816360473633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,2,power_law_1.01,0.8202048301696777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,2,power_law_1.01,1.6032960891723633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,2,power_law_1.01,1.1215935707092286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,2,power_law_1.01,0.8539903640747071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,2,power_law_1.01,1.1387776374816894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,2,1,power_law_1.2,8.068621063232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,2,balanced,9.509471893310547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,2,power_law_1.2,0.12700799703598023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,2,balanced,0.44283199310302734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,2,power_law_1.2,0.256166410446167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,2,balanced,0.4493013223012288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,2,balanced,0.4657226800918579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,2,power_law_1.2,0.3372159957885742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,2,power_law_1.01,0.8437888145446777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,2,power_law_1.2,0.506604814529419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,2,power_law_1.01,0.897606372833252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,2,power_law_1.01,1.2494784355163575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,2,power_law_1.2,0.6413887977600098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,2,power_law_1.01,0.9171584129333497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,2,power_law_1.01,1.435654354095459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,2,power_law_1.2,0.8782784461975097
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,2,power_law_1.01,0.052070397138595584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,2,power_law_1.01,0.9010880470275879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,2,balanced,12.923535664876303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,2,1,power_law_1.2,10.41607666015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,2,power_law_1.01,0.07525119781494141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,2,power_law_1.01,1.6536127090454102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,2,balanced,0.4726933240890503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,2,power_law_1.01,0.9412287712097168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,2,power_law_1.2,1.1208255767822266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,2,balanced,0.4926826556523641
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,2,power_law_1.01,0.10269440412521362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,2,power_law_1.2,0.10505599975585937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,2,balanced,0.5144160191218058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,2,power_law_1.01,0.9529727935791016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,2,power_law_1.01,0.14556800127029418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,2,power_law_1.2,1.2362943649291993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,2,power_law_1.2,0.17418240308761596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,2,power_law_1.01,1.744044876098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,2,power_law_1.01,0.20579841136932372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,2,power_law_1.01,1.0240575790405273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,2,power_law_1.2,1.33569278717041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,2,power_law_1.01,0.26150400638580323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,2,power_law_1.01,1.052006435394287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,2,power_law_1.01,0.32181758880615235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,2,power_law_1.01,1.6281152725219727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,2,power_law_1.2,1.304742431640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,2,power_law_1.01,0.34535679817199705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,2,power_law_1.2,0.09941120147705078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,2,power_law_1.01,1.2254976272583007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,2,power_law_1.2,0.15716480016708373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,2,power_law_1.2,1.4551551818847657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,2,power_law_1.01,1.3475071907043457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,2,power_law_1.2,0.22450559139251708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,2,power_law_1.2,1.516697597503662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,2,power_law_1.2,0.33022079467773435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,2,power_law_1.01,1.6410432815551759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,2,power_law_1.2,0.47447681427001953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,2,power_law_1.2,1.6084159851074218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,2,power_law_1.2,0.5485504150390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,2,power_law_1.01,1.9272384643554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,2,power_law_1.01,0.34854400157928467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,2,power_law_1.01,2.058803176879883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,2,power_law_1.2,1.6371776580810546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,2,power_law_1.2,0.7793087959289551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,2,power_law_1.2,0.23107199668884276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,2,power_law_1.01,0.3558784008026123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,2,power_law_1.2,0.32490880489349366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,2,power_law_1.01,1.7682304382324219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,2,power_law_1.2,0.8273216247558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,2,power_law_1.2,1.680633544921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,2,power_law_1.01,2.5647104263305662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,2,power_law_1.01,2.4389631271362306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,2,power_law_1.2,0.4531392097473145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,2,power_law_1.2,0.8467519760131836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,2,power_law_1.01,1.9083967208862305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,2,power_law_1.2,1.7379327774047852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,2,power_law_1.2,0.5808703899383545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,2,power_law_1.2,0.8374527931213379
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,2,balanced,0.5411039988199869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,2,power_law_1.01,3.0808576583862304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,2,power_law_1.2,0.7811391830444336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,2,balanced,0.6993439992268881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,2,power_law_1.01,3.3107135772705076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,2,power_law_1.2,0.928281593322754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,2,power_law_1.01,2.130451202392578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,2,balanced,0.6384586493174235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,2,power_law_1.2,0.7738815784454346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,2,power_law_1.01,0.3615744113922119
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,2,balanced,1.1737013657887776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,2,power_law_1.2,0.9993151664733887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,2,power_law_1.01,0.38776960372924807
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,2,balanced,1.046565294265747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,2,power_law_1.2,0.8071743965148925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,2,power_law_1.01,2.4842559814453127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,2,power_law_1.2,1.0410752296447754
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,2,power_law_1.01,0.3972032070159912
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,2,balanced,1.4845387140909831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,2,power_law_1.01,4.083180618286133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,2,power_law_1.2,0.8674367904663086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,2,power_law_1.01,0.4321983814239502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,2,power_law_1.2,1.0874879837036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,2,power_law_1.01,0.4029183864593506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,2,power_law_1.2,0.8805439949035645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,2,power_law_1.2,1.8524608612060547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,2,power_law_1.2,1.1402688026428223
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,2,power_law_1.01,0.4234879970550537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,2,power_law_1.2,0.8876223564147949
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,2,power_law_1.01,0.4669312000274658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,2,power_law_1.2,1.2294976234436035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,2,power_law_1.2,2.0360511779785155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,2,power_law_1.2,0.9300928115844727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,2,power_law_1.01,0.4836415767669678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,2,power_law_1.01,5.619283294677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,2,power_law_1.2,1.2922240257263184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,2,power_law_1.2,0.9509568214416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,2,power_law_1.2,2.2776575088500977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,2,power_law_1.2,0.9915712356567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,2,balanced,1.9210453033447266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,2,power_law_1.2,1.498969554901123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,2,power_law_1.01,2.8928512573242187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,2,power_law_1.2,1.0363200187683106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,2,power_law_1.2,1.7356096267700196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,2,power_law_1.2,2.6283647537231447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,2,power_law_1.2,1.1186559677124024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,2,power_law_1.01,4.372019195556641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,2,power_law_1.01,3.7397823333740234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,2,power_law_1.2,1.2761152267456055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,2,power_law_1.01,7.194464111328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,2,power_law_1.2,2.9963327407836915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,2,power_law_1.2,1.4217984199523925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,2,power_law_1.2,1.7000703811645508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,2,power_law_1.01,4.564108657836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,2,power_law_1.2,3.7413631439208985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,2,power_law_1.01,0.5257023811340332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,2,power_law_1.01,5.708172988891602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,2,power_law_1.2,2.0639680862426757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,2,power_law_1.2,2.006111907958984
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,2,power_law_1.01,0.5562111854553222
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,2,power_law_1.01,0.6280447959899902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,2,power_law_1.2,2.521459197998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,2,power_law_1.01,0.7638591766357422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,2,power_law_1.2,4.805145645141602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,2,power_law_1.01,6.697164916992188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,2,power_law_1.01,0.8363583564758301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,2,power_law_1.2,3.2806015014648438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,2,power_law_1.01,8.034848022460938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,2,power_law_1.01,1.0605888366699219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,2,power_law_1.2,0.05211520195007324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,2,power_law_1.2,0.07514240145683289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,2,power_law_1.01,1.3536576271057128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,2,power_law_1.2,0.09579520225524903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,2,power_law_1.2,0.1386944055557251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,2,power_law_1.2,4.225747299194336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,2,power_law_1.2,0.17806719541549682
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,2,power_law_1.01,1.7721664428710937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,2,power_law_1.01,8.06153564453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,2,power_law_1.2,0.22714879512786865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,2,power_law_1.2,0.30560638904571535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,2,power_law_1.2,0.3384255886077881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,2,power_law_1.01,10.278470611572265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,2,power_law_1.2,0.3405247926712036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,2,power_law_1.2,7.220172882080078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,2,power_law_1.01,10.005689239501953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,2,power_law_1.2,5.721388626098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,2,power_law_1.2,0.3458240032196045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,2,1,balanced,0.08469333251317342
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,2,power_law_1.2,0.3680000066757202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,2,power_law_1.2,0.3706239938735962
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,2,power_law_1.2,0.414137601852417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,2,power_law_1.2,2.664179229736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,2,power_law_1.2,0.4460031986236572
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,2,power_law_1.2,0.41181440353393556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,2,power_law_1.2,0.4395264148712158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,2,1,balanced,0.1264693339665731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,2,power_law_1.2,8.384031677246094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,2,power_law_1.2,0.4902912139892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,2,1,balanced,0.25358400742212933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,2,power_law_1.2,3.16549129486084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,2,1,balanced,0.44899733861287433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,2,power_law_1.2,0.4978367805480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,2,1,balanced,0.8202293713887533
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,2,power_law_1.2,0.566700792312622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,2,1,balanced,1.5848320325215657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,2,power_law_1.01,3.158060836791992
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,2,power_law_1.2,0.5906176090240478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,2,power_law_1.01,13.343212890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,2,1,balanced,1.5903466542561848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,2,1,balanced,0.0786186655362447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,2,1,balanced,0.09960533181826274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,2,power_law_1.2,0.6644608020782471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,2,power_law_1.2,4.524428939819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,2,1,balanced,1.5903253555297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,2,power_law_1.2,7.59912338256836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,2,power_law_1.2,0.8072128295898438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,2,1,balanced,1.599071979522705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,2,power_law_1.2,0.9082559585571289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,2,1,balanced,1.6113920211791992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,2,1,balanced,1.614672025044759
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,2,power_law_1.2,1.1877951622009277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,2,1,balanced,0.07284266750017802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,2,1,balanced,0.0943946639696757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,2,1,balanced,0.1693013310432434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,2,power_law_1.2,5.6833984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,2,1,balanced,0.1602186659971873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,2,1,balanced,0.2866293390591939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,2,power_law_1.2,12.310598754882813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,2,1,balanced,0.28991466760635376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,2,1,balanced,0.545685331026713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,2,power_law_1.2,1.425107192993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,2,power_law_1.01,12.726886749267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,2,1,balanced,1.0453973611195881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,2,1,balanced,1.0472426414489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,2,power_law_1.2,1.8969728469848632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,2,1,balanced,1.049770673116048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,2,1,balanced,1.051418701807658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,2,power_law_1.2,10.635897827148437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,2,1,balanced,1.6265066464742024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,2,1,balanced,1.0552373727162678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,2,1,balanced,0.5297439893086752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,2,1,balanced,1.6161120732625325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,2,1,balanced,1.060314655303955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,2,1,balanced,1.0088640054066975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,2,power_law_1.2,3.341075134277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,2,power_law_1.2,8.331417846679688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,2,1,balanced,1.6317599614461262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,2,1,balanced,1.018191973368327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,2,1,balanced,1.0211573441823323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,2,1,balanced,1.785264015197754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,2,1,balanced,1.025381326675415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,2,1,balanced,1.0295573075612385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,2,1,balanced,1.8030826250712078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,2,1,balanced,1.0388960043589275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,2,1,balanced,1.8291786511739094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,2,1,balanced,1.0481546719868977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,2,1,balanced,1.0564853350321453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,2,1,balanced,1.8886772791544597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,2,1,balanced,1.0743306477864583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,2,power_law_1.2,15.794796752929688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,2,power_law_1.01,15.927436828613281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,2,1,balanced,1.0881760120391846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,2,1,balanced,1.9600480397542317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,2,1,power_law_1.01,0.09095680117607116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,2,1,balanced,1.1057013670603435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,2,1,power_law_1.01,0.07661439776420594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,2,1,power_law_1.01,0.12443519830703735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,2,1,balanced,2.0791573524475098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,2,1,balanced,1.1332746346791585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,2,1,power_law_1.01,0.09727360010147094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,2,power_law_1.2,13.543289184570312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,2,1,power_law_1.01,0.24915199279785155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,2,power_law_1.2,10.59112319946289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,2,1,balanced,1.1881653467814128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,2,1,power_law_1.01,0.17226239442825317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,2,1,power_law_1.01,0.4421696186065674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,2,1,balanced,1.065989335378011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,2,1,balanced,1.2376906871795654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,2,1,power_law_1.01,0.2854655981063843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,2,1,balanced,1.0717546939849854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,2,1,balanced,1.3404800097147624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,2,1,power_law_1.01,0.4227136135101318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,2,1,balanced,1.0844799677530925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,2,1,balanced,1.5034133593241374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,2,1,power_law_1.01,0.5590208053588868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,2,1,balanced,1.1441280047098796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,2,1,balanced,1.163376013437907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,2,1,power_law_1.01,0.7742591857910156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,2,1,balanced,2.2550880114237466
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,2,1,balanced,0.03545066714286804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,2,1,balanced,1.2006773153940837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,2,1,balanced,0.05379733443260193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,2,1,balanced,2.260591983795166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,2,1,power_law_1.01,0.8058112144470215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,2,1,power_law_1.01,0.0705344021320343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,2,1,balanced,1.294922669728597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,2,1,balanced,2.511573314666748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,2,1,power_law_1.01,0.6394303798675537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,2,1,power_law_1.01,0.09214079976081849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,2,1,balanced,1.3694613774617512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,2,1,balanced,3.096970558166504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,2,1,power_law_1.01,0.15934079885482788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,2,1,power_law_1.01,0.8392959594726562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,2,1,balanced,1.5378774007161458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,2,1,balanced,3.57640012105306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,2,1,power_law_1.01,0.2899456024169922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,2,1,power_law_1.01,1.160806369781494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,2,1,balanced,1.7156426111857097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,2,1,power_law_1.01,0.4266496181488037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,2,1,balanced,3.4590988159179688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,2,1,balanced,0.07842666904131572
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,2,1,balanced,0.12426666418711345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,2,1,balanced,0.21799999475479126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,2,1,power_law_1.01,1.2083328247070313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,2,1,balanced,0.4021759827931722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,2,1,balanced,2.652143955230713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,2,1,balanced,0.4049066702524821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,2,1,power_law_1.01,0.8244352340698242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,2,1,balanced,0.4134613275527954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,2,1,power_law_1.01,1.248857593536377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,2,1,balanced,0.41568533579508465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,2,1,balanced,0.4172266721725464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,2,1,power_law_1.01,0.8388223648071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,2,1,balanced,5.202773412068685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,2,1,balanced,0.42209601402282715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,2,1,balanced,3.0427465438842773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,2,1,balanced,0.4270079930623372
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,2,1,balanced,0.43424534797668457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,2,1,power_law_1.01,0.8981247901916504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,2,1,balanced,0.4466293255488078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,2,1,balanced,0.4760213295618693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,2,1,power_law_1.01,0.9675583839416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,2,1,power_law_1.01,0.03184640109539032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,2,1,power_law_1.01,0.5855999946594238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,2,1,balanced,4.627589225769043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,2,1,power_law_1.01,0.05379199981689453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,2,1,balanced,4.3723147710164385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,2,1,power_law_1.01,1.030508804321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,2,1,power_law_1.01,0.7711743831634521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,2,1,power_law_1.01,0.07698559761047363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,2,1,balanced,6.396640141805013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,2,1,power_law_1.01,1.1366656303405762
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,2,1,power_law_1.01,0.12331520318984986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,2,1,power_law_1.01,0.8256640434265137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,2,1,power_law_1.01,1.2735936164855957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,2,1,power_law_1.01,0.17501440048217773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,2,1,power_law_1.01,1.0632384300231934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,2,1,power_law_1.01,0.8445247650146485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,2,1,power_law_1.01,0.23612160682678224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,2,1,power_law_1.01,1.336415958404541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,2,1,balanced,5.70359992980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,2,1,balanced,6.757381439208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,2,1,balanced,0.48473600546518963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,2,1,power_law_1.01,1.090988826751709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,2,1,power_law_1.01,0.8502976417541503
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,2,1,power_law_1.01,0.3084671974182129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,2,1,power_law_1.01,1.4170559883117675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,2,1,power_law_1.01,0.8549440383911133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,2,1,power_law_1.01,1.164998435974121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,2,1,power_law_1.01,1.438041591644287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,2,1,balanced,9.595024108886719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,2,1,power_law_1.01,1.3301055908203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,2,1,power_law_1.01,1.5649151802062988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,2,1,power_law_1.01,1.4904383659362792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,2,1,balanced,8.385237375895182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,2,1,balanced,8.846373240152994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,2,1,balanced,0.5072640180587769
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,2,1,balanced,0.5107306639353434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,2,1,power_law_1.01,1.582323169708252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,2,1,power_law_1.01,0.329036808013916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,2,1,power_law_1.01,1.8029695510864259
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,2,1,power_law_1.01,0.3344320058822632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,2,1,power_law_1.01,0.8947008132934571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,2,1,power_law_1.01,1.6406015396118163
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,2,1,power_law_1.01,0.3417407989501953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,2,1,power_law_1.01,2.127673530578613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,2,1,power_law_1.01,0.9121088027954102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,2,1,power_law_1.01,0.3549312114715576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,2,1,power_law_1.01,1.6925247192382813
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,2,1,power_law_1.01,0.36487679481506347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,2,1,power_law_1.01,0.959654426574707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,2,1,power_law_1.01,0.3813568115234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,2,1,balanced,13.15231959025065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,2,1,balanced,0.5366506576538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,2,1,power_law_1.2,0.0833728015422821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,2,1,power_law_1.01,1.8420352935791016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,2,1,power_law_1.01,0.4097792148590088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,2,1,balanced,11.077733357747396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,2,1,power_law_1.2,0.12480000257492066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,2,1,power_law_1.2,0.07958400249481201
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,2,1,power_law_1.01,0.4419839859008789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,2,1,power_law_1.2,0.24854400157928466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,2,1,power_law_1.2,0.09876480102539062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,2,1,power_law_1.01,0.478988790512085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,2,1,power_law_1.2,0.4115327835083008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,2,1,power_law_1.2,0.16926720142364501
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,2,1,power_law_1.01,0.548851203918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,2,1,power_law_1.2,0.5643455982208252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,2,1,power_law_1.2,0.28972160816192627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,2,1,power_law_1.01,0.5180543899536133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,2,1,power_law_1.01,2.808230400085449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,2,1,power_law_1.2,0.7369791984558105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,2,1,power_law_1.01,0.610265588760376
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,2,1,balanced,0.593290646870931
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,2,1,balanced,0.633296012878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,2,1,power_law_1.01,0.6177536010742187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,2,1,power_law_1.2,1.1260160446166991
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,2,1,balanced,0.8095573584238688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,2,1,power_law_1.2,0.07043200135231018
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,2,1,balanced,0.8095893065134684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,2,1,power_law_1.01,3.4724353790283202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,2,1,power_law_1.2,0.09200000166893005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,2,1,power_law_1.2,1.180959987640381
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,2,1,balanced,1.595647970835368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,2,1,power_law_1.2,0.15962239503860473
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,2,1,balanced,1.1296746730804443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,2,1,power_law_1.2,1.257913589477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,2,1,power_law_1.2,0.2797055959701538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,2,1,power_law_1.01,0.9580927848815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,2,1,balanced,1.5960532824198406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,2,1,power_law_1.2,0.37404799461364746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,2,1,power_law_1.2,1.2846400260925293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,2,1,power_law_1.01,1.0140352249145508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,2,1,power_law_1.01,4.813497543334961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,2,1,power_law_1.01,1.9598655700683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,2,1,balanced,2.0587946573893228
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,2,1,power_law_1.01,0.7223616123199463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,2,1,power_law_1.01,1.0679360389709474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,2,1,power_law_1.2,1.3568511962890626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,2,1,power_law_1.01,0.7795455932617188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,2,1,power_law_1.01,1.1851136207580566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,2,1,power_law_1.2,1.399449634552002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,2,1,power_law_1.01,2.29550724029541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,2,1,power_law_1.2,0.38686718940734866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,2,1,power_law_1.01,0.9230784416198731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,2,1,power_law_1.01,1.312172794342041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,2,1,power_law_1.2,1.4480640411376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,2,1,power_law_1.2,0.5160768032073975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,2,1,power_law_1.01,1.0678655624389648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,2,1,power_law_1.01,2.6815872192382812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,2,1,power_law_1.2,0.5358784198760986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,2,1,power_law_1.2,0.7464704036712646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,2,1,power_law_1.01,1.5609919548034668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,2,1,power_law_1.2,1.6525951385498048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,2,1,power_law_1.2,0.7498623847961425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,2,1,power_law_1.01,1.3279104232788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,2,1,power_law_1.2,0.8084159851074219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,2,1,power_law_1.2,1.592313575744629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,2,1,power_law_1.2,0.8029888153076172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,2,1,power_law_1.2,0.8232319831848145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,2,1,power_law_1.01,3.333932876586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,2,1,power_law_1.01,1.786457633972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,2,1,power_law_1.2,1.698124885559082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,2,1,power_law_1.01,6.165100860595703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,2,1,power_law_1.2,1.7449600219726562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,2,1,power_law_1.01,2.707872009277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,2,1,power_law_1.01,4.052204895019531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,2,balanced,0.03151999910672506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,2,balanced,0.038592000802357994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,2,1,power_law_1.01,1.8239103317260743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,2,balanced,0.05566399792830149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,2,balanced,0.08036266764005025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,2,balanced,0.1320693294207255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,2,balanced,0.13078400492668152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,2,balanced,0.13337066769599915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,2,balanced,0.13363200426101685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,2,1,power_law_1.2,0.8598143577575683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,2,balanced,0.14734933773676553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,2,balanced,0.15038933356602988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,2,1,power_law_1.01,2.324787139892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,2,balanced,0.15320533514022827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,2,balanced,0.15545066197713217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,2,1,power_law_1.2,0.9105728149414063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,2,balanced,0.1567039986451467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,2,1,power_law_1.2,0.033164799213409424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,2,balanced,0.16358400384585062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,2,balanced,0.16873067617416382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,2,balanced,0.17545600732167563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,2,1,power_law_1.2,0.05286399722099304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,2,balanced,0.1851466695467631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,2,1,power_law_1.2,0.9644415855407715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,2,balanced,0.20742400487263998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,2,balanced,0.22748267650604248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,2,1,power_law_1.01,5.5981697082519535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,2,1,power_law_1.2,0.07597439885139465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,2,1,power_law_1.01,2.855878448486328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,2,1,power_law_1.01,8.866329956054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,2,1,power_law_1.2,1.9206911087036134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,2,1,power_law_1.2,1.0599360466003418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,2,1,power_law_1.2,0.1151360034942627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,2,1,power_law_1.2,0.15772160291671752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,2,1,power_law_1.2,0.8246015548706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,2,1,power_law_1.2,1.1666303634643556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,2,1,power_law_1.2,0.22042241096496581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,2,1,power_law_1.2,2.0672191619873046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,2,1,power_law_1.2,0.8276800155639649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,2,1,power_law_1.2,1.0850367546081543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,2,1,power_law_1.2,0.3007296085357666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,2,1,power_law_1.01,3.896428680419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,2,1,power_law_1.2,0.8810239791870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,2,1,power_law_1.2,0.323686408996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,2,1,power_law_1.2,1.1305472373962402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,2,1,power_law_1.2,2.3534080505371096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,2,balanced,0.3062666654586792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,2,1,power_law_1.2,0.33420159816741946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,2,balanced,0.35797866185506183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,2,1,power_law_1.01,7.158048248291015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,2,balanced,0.49793068567911786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,2,1,power_law_1.2,1.2115263938903809
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,2,1,power_law_1.2,0.3389695882797241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,2,balanced,0.6349120140075684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,2,1,power_law_1.2,0.3527168035507202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,2,balanced,0.919648011525472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,2,1,power_law_1.2,1.3767616271972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,2,1,power_law_1.2,2.695392036437988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,2,1,power_law_1.2,0.36885120868682864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,2,1,power_law_1.01,4.951846313476563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,2,balanced,1.2006186644236247
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,2,1,power_law_1.2,0.37650558948516843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,2,1,power_law_1.2,1.561184024810791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,2,balanced,1.7633600234985352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,2,1,power_law_1.2,0.913484764099121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,2,1,power_law_1.2,3.4744319915771484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,2,balanced,2.327221393585205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,2,1,power_law_1.2,1.8753023147583008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,2,balanced,0.030239999294281006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,2,balanced,0.035642666121323906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,2,balanced,0.052229334910710655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,2,balanced,0.07299200197060902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,2,balanced,0.1074186662832896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,2,balanced,0.10553066929181416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,2,balanced,0.10773332913716634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,2,balanced,0.10932266712188721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,2,balanced,0.12265599767367046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,2,balanced,0.12432533502578735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,2,1,power_law_1.2,2.214950370788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,2,balanced,0.12532266974449158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,2,1,power_law_1.01,10.522898864746093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,2,balanced,0.1302880048751831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,2,balanced,0.13372266292572021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,2,1,power_law_1.2,0.41729278564453126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,2,1,power_law_1.2,4.243283081054687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,2,1,power_law_1.2,0.45177597999572755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,2,1,power_law_1.2,0.4908031940460205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,2,1,power_law_1.2,2.860390472412109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,2,1,power_law_1.2,0.9090496063232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,2,1,power_law_1.2,0.574732780456543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,2,1,power_law_1.2,0.9736831665039063
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,2,1,power_law_1.2,0.5522175788879394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,2,1,power_law_1.01,7.021702575683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,2,1,power_law_1.2,0.9914303779602051
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,2,1,power_law_1.2,0.6359744071960449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,2,1,power_law_1.01,11.566681671142579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,2,balanced,0.13707199692726135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,2,balanced,0.142767995595932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,2,balanced,0.1502240002155304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,2,balanced,0.16265599926312765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,2,1,power_law_1.2,1.0338751792907714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,2,balanced,0.18820265928904215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,2,balanced,0.03985599925120672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,2,balanced,0.04399466514587402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,2,balanced,0.05336533486843109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,2,balanced,0.07657599945863088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,2,1,power_law_1.2,1.1033727645874023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,2,1,power_law_1.2,1.242630386352539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,2,1,power_law_1.2,5.717862319946289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,2,1,power_law_1.2,1.3523584365844727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,2,1,power_law_1.2,0.6427711963653564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,2,balanced,0.21451199054718018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,2,balanced,0.30537599325180054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,2,1,power_law_1.01,9.09093780517578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,2,1,power_law_1.2,0.7615808010101318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,2,balanced,0.11306132872899373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,2,1,power_law_1.2,1.621516799926758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,2,balanced,0.1135040024916331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,2,balanced,0.11590933799743652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,2,balanced,0.1163146694501241
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,2,1,power_law_1.2,0.7929024219512939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,2,balanced,0.11716266473134358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,2,balanced,0.1200373371442159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,2,balanced,0.1216159959634145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,2,balanced,0.12407466769218445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,2,balanced,0.12782399853070578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,2,1,power_law_1.2,1.8808128356933593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,2,1,power_law_1.2,0.9433728218078613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,2,balanced,0.13358400265375772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,2,1,power_law_1.01,13.634489440917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,2,balanced,0.1404159963130951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,2,balanced,0.1458080013593038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,2,1,power_law_1.2,7.218220520019531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,2,power_law_1.01,0.03261440098285675
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,2,1,power_law_1.2,1.1016575813293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,2,power_law_1.01,0.048172798752784726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,2,1,power_law_1.2,2.3929920196533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,2,balanced,0.36191999912261963
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,2,balanced,0.02385599911212921
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,2,balanced,0.027658666173617046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,2,balanced,0.5090879996617635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,2,1,power_law_1.2,1.3512255668640136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,2,balanced,0.029626667499542236
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,2,balanced,0.03522133330504099
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,2,balanced,0.055498664577802025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,2,balanced,0.05789866546789805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,2,balanced,0.05961066484451294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,2,balanced,0.06142933170000712
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,2,balanced,0.061290666460990906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,2,balanced,0.0633493314186732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,2,1,power_law_1.2,3.542988967895508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,2,1,power_law_1.2,2.916582489013672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,2,1,power_law_1.2,1.7982336044311524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,2,balanced,0.15602667133013406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,2,balanced,0.17935999234517416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,2,balanced,0.2036799987157186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,2,balanced,0.2950719992319743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,2,balanced,0.34035201867421466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,2,balanced,0.4765333334604899
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,2,1,power_law_1.2,2.878982353210449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,2,1,power_law_1.2,3.947078323364258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,2,1,power_law_1.2,10.364121246337891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,2,power_law_1.01,0.03240320086479187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,2,balanced,0.06503466765085857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,2,balanced,0.062224000692367554
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,2,balanced,0.06274133423964183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,2,balanced,0.06427733103434245
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,2,balanced,0.06467199822266896
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,2,balanced,0.06605866551399231
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,2,balanced,0.06846933563550313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,2,balanced,0.07666666805744171
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,2,balanced,0.07855999966462453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,2,balanced,0.09809066851933797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,2,balanced,0.10338667035102844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,2,balanced,0.1536746621131897
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,2,balanced,0.1622666617234548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,2,balanced,0.6137653191884359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,2,balanced,0.22403200467427573
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,2,balanced,0.2841599980990092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,2,balanced,0.8815893332163492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,2,1,power_law_1.2,4.882540893554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,2,balanced,0.40026132265726727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,2,power_law_1.01,0.04170880019664765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,2,balanced,0.5246293147404989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,2,balanced,1.156272013982137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,2,power_law_1.01,0.05482879877090454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,2,power_law_1.01,0.06598399877548218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,2,balanced,1.6944212913513184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,2,power_law_1.01,0.07382400035858154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,2,power_law_1.01,0.061267197132110596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,2,power_law_1.01,0.071424001455307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,2,1,power_law_1.2,4.997990417480469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,2,balanced,0.6557493209838867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,2,power_law_1.01,0.077920001745224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,2,power_law_1.01,0.11357439756393432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,2,balanced,0.948911984761556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,2,power_law_1.01,0.12437119483947753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,2,1,power_law_1.2,13.708493041992188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,2,power_law_1.01,0.12620799541473388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,2,power_law_1.01,0.14368000030517578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,2,power_law_1.01,0.14481279850006104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,2,power_law_1.01,0.1478976011276245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,2,power_law_1.01,0.09403520226478576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,2,power_law_1.01,0.09673600196838379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,2,power_law_1.01,0.0992576003074646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,2,power_law_1.01,0.1172287940979004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,2,1,power_law_1.2,7.055052947998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,2,1,power_law_1.2,6.230278396606446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,2,balanced,1.2426453431447346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,2,power_law_1.01,0.025766399502754212
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,2,power_law_1.01,0.028870400786399842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,2,balanced,1.8268532752990723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,2,power_law_1.01,0.03706879913806915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,2,power_law_1.01,0.15232640504837036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,2,power_law_1.01,0.04674560129642487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,2,balanced,2.2390400568644204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,2,balanced,2.4145387013753257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,2,power_law_1.2,0.03475840091705322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,2,power_law_1.2,0.03264000117778778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,2,power_law_1.2,0.04625920057296753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,2,power_law_1.2,0.04305280148983002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,2,power_law_1.01,0.11977599859237671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,2,power_law_1.2,0.061273598670959474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,2,power_law_1.01,0.12369920015335083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,2,power_law_1.2,0.05378559827804565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,2,power_law_1.01,0.12743040323257446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,2,power_law_1.2,0.061791998147964475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,2,power_law_1.01,0.13109760284423827
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,2,power_law_1.01,0.0319680005311966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,2,1,power_law_1.2,8.931718444824218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,2,1,power_law_1.2,9.110841369628906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,2,power_law_1.01,0.03548159897327423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,2,power_law_1.01,0.15400320291519165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,2,power_law_1.01,0.04310399889945984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,2,power_law_1.01,0.060083198547363284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,2,power_law_1.01,0.16064640283584594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,2,power_law_1.01,0.05459200143814087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,2,power_law_1.01,0.07395200133323669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,2,power_law_1.01,0.1688447952270508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,2,power_law_1.01,0.05616639852523804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,2,power_law_1.01,0.08150399923324585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,2,power_law_1.01,0.17662080526351928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,2,power_law_1.01,0.05861759781837463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,2,power_law_1.01,0.10071040391921997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,2,power_law_1.01,0.18924800157546998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,2,power_law_1.01,0.05988479852676391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,2,power_law_1.01,0.10814080238342286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,2,power_law_1.2,0.07434880137443542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,2,power_law_1.2,0.07740799784660339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,2,power_law_1.01,0.1403712034225464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,2,power_law_1.2,0.1211135983467102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,2,power_law_1.01,0.15006719827651976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,2,power_law_1.2,0.12522239685058595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,2,power_law_1.01,0.1590208053588867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,2,power_law_1.2,0.12711039781570435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,2,power_law_1.01,0.18117120265960693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,2,power_law_1.2,0.036883199214935304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,2,power_law_1.01,0.21928319931030274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,2,power_law_1.2,0.04655359983444214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,2,power_law_1.01,0.2500992059707642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,2,power_law_1.2,0.053939199447631835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,2,power_law_1.01,0.2328255891799927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,2,power_law_1.01,0.31986560821533205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,2,power_law_1.2,0.06880000233650208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,2,1,power_law_1.2,11.603743743896484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,2,power_law_1.01,0.2632191896438599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,2,power_law_1.2,0.07984640002250672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,2,power_law_1.01,0.404633617401123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,2,power_law_1.01,0.11091840267181396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,2,power_law_1.2,0.06793599724769592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,2,power_law_1.2,0.10183680057525635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,2,power_law_1.01,0.11265920400619507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,2,power_law_1.2,0.09251199960708618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,2,power_law_1.01,0.5715007781982422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,2,power_law_1.2,0.10975359678268433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,2,power_law_1.2,0.09535999894142151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,2,power_law_1.2,0.10943360328674316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,2,power_law_1.2,0.14380799531936644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,2,power_law_1.01,0.7108736038208008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,2,power_law_1.2,0.10335999727249146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,2,power_law_1.2,0.11288319826126099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,2,power_law_1.2,0.14577920436859132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,2,power_law_1.2,0.11901439428329467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,2,power_law_1.2,0.11580159664154052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,2,power_law_1.2,0.14843519926071166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,2,power_law_1.01,1.0194496154785155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,2,power_law_1.2,0.12142080068588257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,2,power_law_1.2,0.1171455979347229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,2,power_law_1.2,0.15374079942703248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,2,power_law_1.2,0.12398079633712769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,2,power_law_1.2,0.15722880363464356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,2,power_law_1.01,1.3436032295227052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,2,power_law_1.2,0.1637887954711914
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,2,power_law_1.2,0.02513279914855957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,2,power_law_1.2,0.16979199647903442
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,2,power_law_1.2,0.027692800760269164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,2,power_law_1.2,0.17683199644088746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,2,power_law_1.2,0.030822399258613586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,2,power_law_1.01,0.1152959942817688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,2,power_law_1.2,0.19279359579086303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,2,power_law_1.01,1.9878143310546874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,2,power_law_1.2,0.034720000624656674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,2,power_law_1.01,0.11749759912490845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,2,power_law_1.2,0.23634560108184816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,2,power_law_1.2,0.04020479917526245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,2,power_law_1.2,0.2686208009719849
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,2,power_law_1.2,0.0526528000831604
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,2,power_law_1.01,0.06072319746017456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,2,power_law_1.2,0.05667200088500977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,2,power_law_1.01,0.06427519917488098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,2,power_law_1.2,0.12309759855270386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,2,power_law_1.01,2.5785152435302736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,2,power_law_1.01,0.06223359704017639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,2,power_law_1.2,0.12513920068740844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,2,power_law_1.01,0.06273919939994813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,2,power_law_1.2,0.13393280506134034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,2,power_law_1.01,0.06581760048866273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,2,power_law_1.2,0.14224640130996705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,2,power_law_1.01,0.06620799899101257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,2,power_law_1.2,0.15571839809417726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,2,power_law_1.01,0.06727679967880248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,2,power_law_1.2,0.17000319957733154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,2,power_law_1.01,0.12078720331192017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,2,power_law_1.01,0.07400959730148315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,2,power_law_1.2,0.34064640998840334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,2,power_law_1.01,0.12327680587768555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,2,power_law_1.01,0.08025599718093872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,2,power_law_1.01,0.133024001121521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,2,power_law_1.01,0.09066240191459655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,2,1,balanced,0.027317332724730175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,2,power_law_1.01,0.33393919467926025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,2,power_law_1.01,0.14216320514678954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,2,power_law_1.01,0.11060479879379273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,2,power_law_1.01,0.40597758293151853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,2,power_law_1.01,0.15459200143814086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,2,power_law_1.01,0.12535680532455445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,2,power_law_1.01,0.16462080478668212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,2,power_law_1.01,0.5570240020751953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,2,power_law_1.01,0.20317440032958983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,2,power_law_1.01,0.7093183994293213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,2,power_law_1.01,0.24119040966033936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,2,power_law_1.2,0.20960640907287598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,2,power_law_1.01,0.9908479690551758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,2,power_law_1.2,0.2369152069091797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,2,power_law_1.2,0.05904639959335327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,2,power_law_1.2,0.1282047986984253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,2,1,balanced,0.033941333492596946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,2,power_law_1.2,0.060224002599716185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,2,power_law_1.01,1.309235191345215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,2,power_law_1.2,0.06079999804496765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,2,power_law_1.2,0.06440320014953613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,2,power_law_1.01,0.15484800338745117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,2,power_law_1.2,0.06155520081520081
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,2,power_law_1.01,0.18889600038528442
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,2,power_law_1.2,0.06325119733810425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,2,power_law_1.01,0.2566272020339966
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,2,power_law_1.2,0.0669376015663147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,2,power_law_1.01,0.3117311954498291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,2,power_law_1.01,0.3141247987747192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,2,power_law_1.01,0.39001600742340087
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,2,power_law_1.01,0.45453438758850095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,2,power_law_1.2,0.3137792110443115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,2,1,balanced,0.04975999891757965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,2,power_law_1.01,0.5211391925811768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,2,power_law_1.2,0.1345088005065918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,2,power_law_1.2,0.38528640270233155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,2,power_law_1.01,0.5844287872314453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,2,power_law_1.2,0.1420032024383545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,2,1,balanced,0.025392000873883564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,2,1,balanced,0.028789333999156952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,2,power_law_1.01,0.6685056209564209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,2,1,balanced,0.044250667095184326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,2,power_law_1.2,0.15184639692306517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,2,1,balanced,0.07180800040562947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,2,1,balanced,0.10900800426801045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,2,1,balanced,0.1074186662832896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,2,power_law_1.2,0.16353280544281007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,2,power_law_1.01,1.929529571533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,2,power_law_1.01,0.9325375556945801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,2,power_law_1.2,0.18133120536804198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,2,power_law_1.2,0.21765758991241455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,2,power_law_1.2,0.06663680076599121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,2,power_law_1.01,1.2636159896850585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,2,power_law_1.2,0.06726400256156921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,2,power_law_1.01,2.4763391494750975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,2,power_law_1.2,0.07340800166130065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,2,1,balanced,0.07471466561158498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,2,1,balanced,0.12984533111254373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,2,power_law_1.2,0.08175359964370728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,2,1,balanced,0.13090667128562927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,2,1,balanced,0.1309333344300588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,2,power_law_1.01,1.7609983444213868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,2,1,balanced,0.13160533706347147
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,2,power_law_1.2,0.08856319785118102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,2,1,balanced,0.13266133268674216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,2,1,balanced,0.13290133078893027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,2,power_law_1.2,0.5274432182312012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,2,1,balanced,0.029418667157491047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,2,1,balanced,0.1330560048421224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,2,power_law_1.2,0.11541759967803955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,2,1,balanced,0.15096533298492432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,2,1,balanced,0.10935999949773152
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,2,power_law_1.2,0.12954879999160768
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,2,power_law_1.2,0.1647487998008728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,2,power_law_1.01,2.3903871536254884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,2,power_law_1.2,0.25292799472808836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,2,power_law_1.2,0.32999041080474856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,2,power_law_1.2,0.41936001777648924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,2,power_law_1.2,0.4117311954498291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,2,power_law_1.2,0.6965184211730957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,2,1,balanced,0.03133866687615713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,2,1,balanced,0.03967999915281931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,2,1,balanced,0.1492639978726705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,2,1,balanced,0.10954667131106059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,2,1,balanced,0.15096533298492432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,2,1,balanced,0.109333336353302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,2,1,balanced,0.1543786625067393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,2,1,balanced,0.11200533310572307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,2,1,balanced,0.15871999661127725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,2,1,balanced,0.11199466387430827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,2,1,balanced,0.16523200273513794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,2,1,balanced,0.12312533458073933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,2,1,balanced,0.1814346710840861
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,2,1,balanced,0.02521066615978877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,2,1,balanced,0.1967680056889852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,2,1,balanced,0.2558986743291219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,2,1,balanced,0.29135467608769733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,2,1,balanced,0.39952532450358075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,2,1,balanced,0.5066719849904379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,2,1,balanced,0.7280480066935221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,2,1,balanced,0.9461013476053873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,2,1,balanced,0.06217599908510844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,2,1,balanced,0.09628799557685852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,2,1,balanced,0.09525866309801738
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,2,power_law_1.2,0.19723520278930665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,2,1,balanced,0.09622400005658467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,2,power_law_1.2,0.5599679946899414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,2,1,balanced,0.125301331281662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,2,1,balanced,0.026341333985328674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,2,1,balanced,0.02972800036271413
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,2,1,balanced,0.03676266719897588
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,2,1,balanced,0.056101332108179726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,2,1,balanced,0.0581226646900177
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,2,1,balanced,0.05937066674232483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,2,1,balanced,0.06102933486302694
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,2,1,balanced,0.06243200103441874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,2,1,balanced,0.06440000236034393
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,2,1,balanced,0.065610667069753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,2,1,balanced,0.06858666737874348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,2,1,balanced,0.07056533296902974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,2,1,balanced,0.0724426656961441
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,2,1,balanced,0.07069866855939229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,2,1,balanced,0.07261333366235097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,2,power_law_1.2,0.5636159896850585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,2,1,balanced,0.07793599863847096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,2,1,balanced,0.0976639986038208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,2,1,balanced,0.08327466746171315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,2,1,balanced,0.09934399525324504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,2,1,balanced,0.08919999996821086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,2,1,balanced,0.09971200426419576
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,2,1,balanced,0.10875733693440755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,2,1,balanced,0.10110933581988017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,2,1,balanced,0.11372799674669902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,2,1,balanced,0.10366400082906087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,2,1,balanced,0.10521066188812256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,2,power_law_1.2,0.9514240264892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,2,1,balanced,0.10956266522407532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,2,power_law_1.2,0.7671743869781494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,2,1,balanced,0.137061337629954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,2,1,balanced,0.1392159958680471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,2,1,balanced,0.14403200149536133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,2,1,balanced,0.1567520002524058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,2,1,balanced,0.17093332608540854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,2,1,balanced,0.2540213267008464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,2,1,balanced,0.28325867652893066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,2,1,balanced,0.39055999120076496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,2,1,balanced,0.5077919960021973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,2,1,balanced,1.3891626993815105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,2,1,balanced,0.7291200160980225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,2,1,balanced,0.12813333670298258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,2,1,balanced,0.9579412937164307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,2,1,balanced,0.19857599337895712
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,2,1,balanced,0.18456000089645386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,2,1,balanced,0.25727999210357666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,2,1,balanced,1.4057226181030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,2,1,balanced,0.3243946631749471
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,2,1,balanced,0.453818678855896
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,2,1,balanced,0.5908746719360352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,2,1,balanced,1.862544059753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,2,1,power_law_1.01,0.021715199947357176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,2,1,power_law_1.01,0.024556800723075867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,2,1,power_law_1.01,0.03319680094718933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,2,1,power_law_1.01,0.046911999583244324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,2,power_law_1.2,0.7252543926239013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,2,1,balanced,0.13331199685732523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,2,1,balanced,0.13834666212399802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,2,1,balanced,0.14456533392270407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,2,1,balanced,0.16290666659673056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,2,1,balanced,0.18331199884414673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,2,1,balanced,0.258842666943868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,2,1,balanced,0.2940640052159627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,2,1,balanced,0.41069332758585614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,2,1,balanced,0.522709329922994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,2,power_law_1.2,1.2917823791503906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,2,power_law_1.2,1.0559231758117675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,2,1,balanced,0.750170628229777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,2,1,power_law_1.01,0.03383040130138397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,2,1,balanced,0.9779146512349447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,2,1,power_law_1.01,0.043635201454162595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,2,1,power_law_1.01,0.06030719876289368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,2,1,power_law_1.01,0.06439679861068726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,2,1,balanced,1.431125322977702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,2,1,power_law_1.01,0.07111039757728577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,2,1,power_law_1.01,0.07794560194015503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,2,1,power_law_1.01,0.09429759979248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,2,1,power_law_1.01,0.12101119756698608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,2,1,balanced,1.8944533665974934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,2,1,power_law_1.01,0.12205439805984497
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,2,power_law_1.2,0.25813119411468505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,2,1,power_law_1.01,0.028652799129486085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,2,1,power_law_1.01,0.03194240033626557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,2,1,power_law_1.01,0.04092159867286682
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,2,power_law_1.2,0.321126389503479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,2,1,power_law_1.01,0.09608960151672363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,2,1,power_law_1.01,0.09886720180511474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,2,1,power_law_1.01,0.12751359939575196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,2,1,power_law_1.01,0.10499199628829955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,2,1,power_law_1.01,0.1259328007698059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,2,1,power_law_1.01,0.11235840320587158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,2,1,power_law_1.01,0.12794239521026612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,2,1,power_law_1.01,0.11980799436569214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,2,1,power_law_1.01,0.13619199991226197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,2,1,power_law_1.01,0.1217344045639038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,2,1,power_law_1.01,0.14460159540176393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,2,1,balanced,1.8325600624084473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,2,1,power_law_1.01,0.05324800014495849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,2,1,power_law_1.01,0.12352639436721802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,2,1,power_law_1.01,0.14627840518951415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,2,1,power_law_1.01,0.06178560256958008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,2,1,power_law_1.01,0.12951040267944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,2,1,power_law_1.01,0.14985599517822265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,2,1,power_law_1.01,0.08684800267219543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,2,1,power_law_1.01,0.13701759576797484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,2,1,power_law_1.01,0.1533504009246826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,2,1,power_law_1.01,0.08885759711265565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,2,1,power_law_1.01,0.14238719940185546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,2,1,power_law_1.01,0.1583680033683777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,2,1,power_law_1.01,0.0899071991443634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,2,1,power_law_1.01,0.15800319910049437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,2,1,power_law_1.01,0.16492799520492554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,2,1,power_law_1.01,0.18645119667053223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,2,1,power_law_1.01,0.20468480587005616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,2,1,power_law_1.01,0.21599359512329103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,2,1,power_law_1.01,0.22355198860168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,2,1,power_law_1.01,0.2712575912475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,2,1,power_law_1.01,0.27273600101470946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,2,1,power_law_1.01,0.32889599800109864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,2,1,power_law_1.01,0.32756481170654295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,2,1,power_law_1.01,0.4431935787200928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,2,1,power_law_1.01,0.43873281478881837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,2,1,power_law_1.01,0.5558591842651367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,2,1,power_law_1.01,0.5471295833587646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,2,1,power_law_1.01,0.0934719979763031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,2,1,power_law_1.01,0.788479995727539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,2,1,power_law_1.01,0.09609599709510804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,2,1,power_law_1.01,0.775923204421997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,2,1,power_law_1.01,0.10216959714889526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,2,1,power_law_1.01,1.017484760284424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,2,1,power_law_1.01,0.11023999452590942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,2,power_law_1.2,1.051251220703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,2,1,power_law_1.01,0.11514879465103149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,2,1,power_law_1.01,0.11636480093002319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,2,1,power_law_1.01,1.4812288284301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,2,1,power_law_1.01,0.12748160362243652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,2,1,power_law_1.01,0.14039039611816406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,2,1,power_law_1.01,0.15278079509735107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,2,1,power_law_1.01,1.9472896575927734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,2,1,power_law_1.01,0.1748095989227295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,2,1,power_law_1.2,0.022124800086021423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,2,1,power_law_1.2,0.0335999995470047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,2,1,power_law_1.01,0.024659200012683867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,2,1,power_law_1.2,0.045228800177574156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,2,1,power_law_1.01,0.026047998666763307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,2,1,power_law_1.01,0.03030399978160858
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,2,1,power_law_1.01,0.03363839983940124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,2,1,power_law_1.01,0.03958399891853333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,2,1,power_law_1.01,0.0514303982257843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,2,1,power_law_1.01,0.05642240047454834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,2,1,power_law_1.01,0.20741119384765624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,2,1,power_law_1.01,0.058208000659942624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,2,1,power_law_1.01,0.2592704057693481
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,2,1,power_law_1.01,0.05902720093727112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,2,1,power_law_1.01,0.9978240013122559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,2,1,power_law_1.01,0.31351680755615235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,2,1,power_law_1.2,0.062412798404693604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,2,1,power_law_1.2,0.07187839746475219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,2,1,power_law_1.01,0.4257728099822998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,2,1,power_law_1.01,1.4534720420837401
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,2,1,power_law_1.2,0.11451519727706909
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,2,power_law_1.2,1.362342357635498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,2,1,power_law_1.01,0.5374144077301025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,2,power_law_1.2,1.3202879905700684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,2,1,power_law_1.01,0.759116792678833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,2,1,power_law_1.01,1.9053119659423827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,2,1,power_law_1.2,0.023948800563812257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,2,1,power_law_1.01,0.981817626953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,2,1,power_law_1.2,0.03418239951133728
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,2,1,power_law_1.01,0.059443199634552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,2,1,power_law_1.2,0.04259839951992035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,2,1,power_law_1.01,0.06290559768676758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,2,1,power_law_1.01,1.4293824195861817
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,2,1,power_law_1.01,0.06466559767723083
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,2,1,power_law_1.01,0.06799359917640686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,2,1,power_law_1.2,0.12049920558929443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,2,1,power_law_1.01,0.0738752007484436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,2,1,power_law_1.2,0.1241152048110962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,2,1,power_law_1.01,1.8711872100830078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,2,1,power_law_1.2,0.1259775996208191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,2,1,power_law_1.2,0.127455997467041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,2,1,power_law_1.2,0.14137599468231202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,2,1,power_law_1.2,0.14476159811019898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,2,1,power_law_1.2,0.028601598739624024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,2,1,power_law_1.2,0.14615039825439452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,2,1,power_law_1.2,0.0318015992641449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,2,1,power_law_1.2,0.05704960227012634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,2,1,power_law_1.2,0.15187840461730956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,2,1,power_law_1.2,0.04039680063724518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,2,1,power_law_1.2,0.06305919885635376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,2,1,power_law_1.2,0.1554304003715515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,2,1,power_law_1.2,0.052172797918319705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,2,1,power_law_1.2,0.0906112015247345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,2,1,power_law_1.2,0.15984640121459961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,2,1,power_law_1.2,0.059308797121047974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,2,1,power_law_1.2,0.09428480267524719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,2,1,power_law_1.2,0.08588799834251404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,2,1,power_law_1.2,0.10037120580673217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,2,1,power_law_1.2,0.08872320055961609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,2,1,power_law_1.2,0.10773119926452637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,2,1,power_law_1.2,0.09159680008888245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,2,1,power_law_1.2,0.11203199625015259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,2,1,power_law_1.2,0.1210752010345459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,2,1,power_law_1.2,0.12058240175247192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,2,1,power_law_1.2,0.1246783971786499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,2,1,power_law_1.2,0.131769597530365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,2,1,power_law_1.2,0.16943999528884887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,2,1,power_law_1.2,0.1390720009803772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,2,1,power_law_1.2,0.2039167881011963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,2,1,power_law_1.2,0.14901119470596313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,2,1,power_law_1.2,0.22576639652252198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,2,1,power_law_1.2,0.1605247974395752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,2,1,power_law_1.2,0.0942911982536316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,2,1,power_law_1.2,0.1906048059463501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,2,1,power_law_1.2,0.09660159945487976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,2,1,power_law_1.2,0.21800320148468016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,2,1,power_law_1.2,0.10273280143737792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,2,1,power_law_1.2,0.2744640111923218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,2,1,power_law_1.2,0.11075199842453003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,2,1,power_law_1.2,0.11745920181274414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,2,1,power_law_1.01,0.06854400038719177
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,2,1,power_law_1.01,0.071424001455307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,2,1,power_law_1.01,0.07914239764213563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,2,1,power_law_1.01,0.08417919874191285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,2,1,power_law_1.2,0.2787584066390991
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,2,1,power_law_1.01,0.09355520009994507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,2,1,power_law_1.2,0.3336384057998657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,2,1,power_law_1.01,0.1027008056640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,2,1,power_law_1.01,0.12007039785385132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,2,1,power_law_1.2,0.33048319816589355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,2,1,power_law_1.01,0.16728960275650023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,2,1,power_law_1.2,0.1175104022026062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,2,1,power_law_1.01,0.20458879470825195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,2,1,power_law_1.2,0.12842880487442015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,2,1,power_law_1.01,0.2712640047073364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,2,1,power_law_1.2,0.1440832018852234
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,2,1,power_law_1.01,0.3390336036682129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,2,1,power_law_1.2,0.1589951992034912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,2,1,power_law_1.2,0.17942399978637696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,2,1,power_law_1.01,0.46459522247314455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,2,1,power_law_1.2,0.20878078937530517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,2,1,power_law_1.2,0.25973119735717776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,2,1,power_law_1.01,0.6048511981964111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,2,1,power_law_1.2,0.31765758991241455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,2,1,power_law_1.2,0.42828798294067383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,2,1,power_law_1.2,0.44659838676452634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,2,1,power_law_1.2,0.5623871803283691
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,2,1,power_law_1.2,0.024537600576877594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,2,1,power_law_1.2,0.44037761688232424
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,2,1,power_law_1.2,0.026316800713539125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,2,1,power_law_1.2,0.029817599058151244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,2,1,power_law_1.2,0.549241590499878
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,2,1,power_law_1.2,0.03242239952087402
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,2,1,power_law_1.2,0.039340800046920775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,2,1,power_law_1.2,0.051609599590301515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,2,1,power_law_1.2,0.05553920269012451
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,2,1,power_law_1.2,0.05826560258865356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,2,1,power_law_1.2,0.058713597059249875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,2,1,power_law_1.2,0.06106240153312683
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,2,1,power_law_1.2,0.06289920210838318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,2,1,power_law_1.2,0.7811647891998291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,2,1,power_law_1.2,0.06513919830322265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,2,1,power_law_1.2,0.06754559874534607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,2,1,power_law_1.2,0.5390528202056885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,2,1,power_law_1.2,0.07534080147743225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,2,1,power_law_1.2,0.06847360134124755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,2,1,power_law_1.2,0.07145599722862243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,2,1,power_law_1.2,0.7911039829254151
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,2,1,power_law_1.2,0.07901440262794494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,2,1,power_law_1.2,0.0843392014503479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,2,power_law_1.2,1.7598400115966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,2,1,power_law_1.2,0.09491199851036072
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,2,1,power_law_1.2,0.10373760461807251
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,2,1,power_law_1.2,0.12164479494094849
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,2,1,power_law_1.2,0.1701696038246155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,2,1,power_law_1.2,0.7610112190246582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,2,1,power_law_1.2,0.20502400398254395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,2,1,power_law_1.2,0.2720576047897339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,2,power_law_1.2,1.9844095230102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,2,1,power_law_1.2,0.9868032455444335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,2,1,power_law_1.2,1.0060352325439452
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,2,power_law_1.2,0.4730559825897217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,2,1,power_law_1.2,1.0271807670593263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,2,1,power_law_1.2,0.3401599884033203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,2,power_law_1.2,2.416371154785156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,2,balanced,0.06820799907048543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,2,balanced,0.08983467022577922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,2,balanced,0.1399626632531484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,2,balanced,0.24622933069864908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,2,balanced,0.4426613251368205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,2,balanced,0.44282134373982746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,2,balanced,0.44365866978963214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,2,balanced,0.44466133912404376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,2,balanced,0.45972267786661786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,2,balanced,0.4617600043614705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,2,balanced,0.46902934710184735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,2,balanced,0.4816533327102661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,2,balanced,0.4832586844762166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,2,balanced,0.49376531442006427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,2,balanced,0.5070773363113403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,2,balanced,0.5336693525314331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,2,balanced,0.5481280088424683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,2,1,power_law_1.2,1.4309311866760255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,2,balanced,0.5953439871470133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,2,balanced,0.6599466800689697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,2,balanced,0.9195040067036947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,2,balanced,1.041093349456787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,2,balanced,1.4716854095458984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,2,1,power_law_1.2,1.4826751708984376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,2,balanced,1.8861600557963054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,2,power_law_1.2,2.0019264221191406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,2,balanced,2.755525271097819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,2,balanced,3.600127855936686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,2,power_law_1.2,0.6343423843383789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,2,balanced,0.05217066903909048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,2,balanced,0.07704533139864604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,2,balanced,0.11620266238848369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,2,balanced,0.1729653278986613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,2,balanced,0.29250667492548627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,2,balanced,0.29553600152333576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,2,balanced,0.29737067222595215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,2,balanced,0.2999200026194255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,2,balanced,0.3378613392512004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,2,1,power_law_1.2,1.8824192047119142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,2,balanced,0.343178669611613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,2,balanced,0.3514293432235718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,2,balanced,0.35655999183654785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,2,balanced,0.36748798688252765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,2,balanced,0.3834666808446248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,2,balanced,0.06062933305899302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,2,balanced,0.07218133409818013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,2,balanced,0.4018293221791585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,2,balanced,5.790314356486003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,2,balanced,0.41862932840983075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,2,balanced,0.442197322845459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,2,balanced,0.5047200123469034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,2,balanced,0.570192019144694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,2,balanced,0.8628693421681722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,2,balanced,1.0051466623942058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,2,balanced,1.4372800191243489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,2,balanced,0.1111946702003479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,2,balanced,0.16751466194788614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,2,balanced,0.2988213300704956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,2,balanced,1.872437318166097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,2,balanced,0.3001599907875061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,2,balanced,0.30471465984980267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,2,balanced,0.30736533800760907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,2,balanced,0.3115413387616475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,2,balanced,0.3141813278198242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,2,balanced,2.7386719385782876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,2,balanced,0.31940267483393353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,2,balanced,0.32767999172210693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,2,balanced,0.3359786669413249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,2,balanced,0.3460533221562703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,2,balanced,0.35941867033640545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,2,balanced,0.37169599533081055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,2,balanced,3.6058079401652017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,2,balanced,0.39898133277893066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,2,balanced,8.37050692240397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,2,balanced,0.44646934668223065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,2,balanced,0.4994613329569499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,2,balanced,0.7385546366373698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,2,balanced,0.8406559626261393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,2,balanced,1.2004746596018474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,2,balanced,1.5567520459493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,2,balanced,2.274149258931478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,2,balanced,2.9882987340291343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,2,balanced,5.334330876668294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,2,balanced,0.027727998793125153
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,2,balanced,0.032511999209721885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,2,balanced,0.05459733307361603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,2,balanced,4.406154632568359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,2,balanced,0.07952000200748444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,2,balanced,0.12386666735013326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,2,balanced,0.12812266747156778
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,2,balanced,0.12981866796811423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,2,balanced,0.1339946687221527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,2,balanced,0.13573333621025085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,2,balanced,0.13838932911554971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,2,balanced,7.062309265136719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,2,balanced,0.14220266540845236
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,2,balanced,0.13875200351079306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,2,balanced,0.14220266540845236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,2,power_law_1.01,0.07984640002250672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,2,balanced,0.14613333344459534
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,2,balanced,0.14616533120473227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,2,balanced,0.14841066797574362
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,2,1,power_law_1.2,0.466323184967041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,2,balanced,0.1547040045261383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,2,balanced,0.16400532921155295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,2,balanced,0.172106663386027
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,2,balanced,0.2394933303197225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,2,balanced,0.20357332626978555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,2,balanced,0.3802666664123535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,2,balanced,0.3161440094312032
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,2,balanced,0.445029338200887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,2,balanced,0.5754719972610474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,2,balanced,0.8316160043080648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,2,power_law_1.01,0.12774399518966675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,2,balanced,5.822682698567708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,2,balanced,1.092293341954549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,2,power_law_1.01,0.17327359914779664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,2,power_law_1.01,0.06992639899253845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,2,power_law_1.01,0.21684479713439941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,2,power_law_1.01,0.09095680117607116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,2,power_law_1.01,0.2818495988845825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,2,power_law_1.01,0.12239999771118164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,2,power_law_1.01,0.06650239825248719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,2,power_law_1.01,0.3854207992553711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,2,power_law_1.01,0.15809919834136962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,2,power_law_1.01,0.08540800213813782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,2,power_law_1.01,0.3899327993392944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,2,power_law_1.01,0.1928447961807251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,2,power_law_1.01,0.11816960573196411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,2,power_law_1.01,0.25217280387878416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,2,power_law_1.01,0.15548800230026244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,2,power_law_1.01,0.26004478931427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,2,power_law_1.01,0.2010495901107788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,2,power_law_1.01,0.285696005821228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,2,power_law_1.01,0.2615231990814209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,2,power_law_1.01,0.3318016052246094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,2,power_law_1.01,0.27244160175323484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,2,power_law_1.01,0.3322688102722168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,2,power_law_1.01,0.27589759826660154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,2,power_law_1.01,0.34495999813079836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,2,power_law_1.01,0.2863424062728882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,2,power_law_1.01,0.35844480991363525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,2,power_law_1.01,0.30150399208068845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,2,power_law_1.01,0.3774271965026855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,2,power_law_1.01,0.30884480476379395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,2,power_law_1.01,0.44213118553161623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,2,power_law_1.01,0.31381759643554685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,2,power_law_1.01,0.4445824146270752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,2,power_law_1.01,0.3217344045639038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,2,power_law_1.01,0.452620792388916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,2,power_law_1.01,0.3463040113449097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,2,power_law_1.01,0.46767358779907225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,2,power_law_1.01,0.3592063903808594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,2,power_law_1.01,0.46943359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,2,power_law_1.01,0.38324480056762694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,2,power_law_1.01,0.5212800025939941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,2,power_law_1.01,0.3655424118041992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,2,power_law_1.01,0.43033599853515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,2,power_law_1.2,2.651276779174805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,2,power_law_1.01,0.3920000076293945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,2,power_law_1.01,0.5198400020599365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,2,power_law_1.01,0.4268479824066162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,2,power_law_1.01,0.608518409729004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,2,power_law_1.01,0.4508927822113037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,2,power_law_1.01,0.5146495819091796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,2,power_law_1.01,0.813593578338623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,2,power_law_1.01,0.6327936172485351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,2,power_law_1.01,0.9571647644042969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,2,power_law_1.01,0.747379207611084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,2,power_law_1.01,0.547270393371582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,2,power_law_1.01,1.3186495780944825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,2,power_law_1.01,0.9368831634521484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,2,power_law_1.01,0.5611392021179199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,2,power_law_1.01,0.5966400146484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,2,power_law_1.01,1.1527744293212892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,2,1,power_law_1.2,1.4505279541015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,2,power_law_1.01,0.6994880199432373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,2,power_law_1.01,1.6856895446777345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,2,power_law_1.01,0.799891185760498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,2,power_law_1.01,0.9948224067687989
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,2,power_law_1.01,0.03333759903907776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,2,power_law_1.01,2.192947196960449
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,2,power_law_1.01,0.04789760112762451
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,2,power_law_1.01,0.05720959901809693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,2,power_law_1.01,1.6953920364379882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,2,power_law_1.01,0.07085440158843995
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,2,power_law_1.01,0.08267520070075988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,2,power_law_1.01,3.094316864013672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,2,power_law_1.01,0.11475199460983276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,2,power_law_1.01,0.11880960464477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,2,power_law_1.01,2.574003219604492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,2,power_law_1.01,0.12420480251312256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,2,power_law_1.01,0.12616319656372071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,2,power_law_1.01,1.214252758026123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,2,power_law_1.01,0.13082879781723022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,2,power_law_1.01,3.8847423553466798
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,2,power_law_1.01,0.14039039611816406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,2,power_law_1.01,3.2740097045898438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,2,power_law_1.01,1.6366144180297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,2,power_law_1.01,2.1071039199829102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,2,power_law_1.01,5.967033767700196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,2,power_law_1.01,4.760441589355469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,2,power_law_1.01,0.1295040011405945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,2,power_law_1.01,3.118118476867676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,2,power_law_1.01,0.13803520202636718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,2,power_law_1.01,0.14908159971237184
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,2,power_law_1.01,0.14870400428771974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,2,power_law_1.01,0.15589760541915892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,2,power_law_1.01,0.165555202960968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,2,power_law_1.01,4.110688018798828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,2,power_law_1.01,0.18227839469909668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,2,power_law_1.01,0.20113279819488525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,2,power_law_1.01,6.511321258544922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,2,power_law_1.01,7.515353393554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,2,power_law_1.2,0.08053119778633118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,2,power_law_1.2,0.07004160284996033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,2,power_law_1.2,0.10773119926452637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,2,power_law_1.2,0.09197440147399902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,2,power_law_1.2,0.16257280111312866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,2,power_law_1.2,0.12168960571289063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,2,power_law_1.2,0.2109055995941162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,2,power_law_1.2,0.1363584041595459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,2,power_law_1.2,0.23758080005645751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,2,power_law_1.01,5.8507648468017575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,2,power_law_1.2,0.17982720136642455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,2,power_law_1.2,0.35117440223693847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,2,power_law_1.2,0.25338239669799806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,2,power_law_1.2,0.39512319564819337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,2,power_law_1.2,0.26625919342041016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,2,power_law_1.01,0.24335999488830568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,2,power_law_1.2,0.4115583896636963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,2,power_law_1.2,0.2763200044631958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,2,power_law_1.01,0.2525504112243652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,2,power_law_1.2,0.3246783971786499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,2,power_law_1.2,0.4400191783905029
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,2,power_law_1.01,0.32320640087127683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,2,power_law_1.2,0.335315203666687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,2,power_law_1.2,0.4519487857818604
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,2,power_law_1.01,0.42026238441467284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,2,power_law_1.2,0.4599487781524658
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,2,power_law_1.01,0.5101056098937988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,2,power_law_1.2,0.46433281898498535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,2,power_law_1.2,0.4819327831268311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,2,power_law_1.2,0.5376063823699951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,2,power_law_1.2,0.5552576065063477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,2,power_law_1.2,0.5691904067993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,2,power_law_1.2,0.34949119091033937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,2,power_law_1.2,0.35757439136505126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,2,power_law_1.2,0.3719167947769165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,2,power_law_1.01,0.6541888236999511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,2,power_law_1.2,0.4005311965942383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,2,power_law_1.2,0.4392831802368164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,2,power_law_1.01,0.9539775848388672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,2,power_law_1.2,0.4675327777862549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,2,power_law_1.2,0.5142784118652344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,2,power_law_1.01,1.5170751571655274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,2,power_law_1.2,0.6298816204071045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,2,power_law_1.2,0.6134463787078858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,2,power_law_1.2,0.7475840091705322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,2,power_law_1.2,0.727180814743042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,2,power_law_1.01,7.824838256835937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,2,power_law_1.2,0.06702079772949218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,2,power_law_1.2,0.9668479919433594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,2,power_law_1.2,0.8430463790893554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,2,power_law_1.2,0.08593279719352723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,2,power_law_1.2,0.11234560012817382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,2,power_law_1.2,1.2083392143249512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,2,power_law_1.2,0.1575808048248291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,2,power_law_1.2,1.0111488342285155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,2,power_law_1.2,0.17581440210342408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,2,power_law_1.2,0.25428481101989747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,2,power_law_1.2,1.2311871528625489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,2,power_law_1.2,0.27039361000061035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,2,power_law_1.2,0.2818624019622803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,2,power_law_1.2,0.2804352045059204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,2,power_law_1.2,0.3009344100952148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,2,power_law_1.2,0.30503039360046386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,2,power_law_1.2,1.7294336318969727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,2,power_law_1.2,1.7341951370239257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,2,power_law_1.2,0.3216576099395752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,2,1,power_law_1.2,1.9061952590942384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,2,1,balanced,0.04728533327579498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,2,1,balanced,0.07127999762694041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,2,1,balanced,0.1232319970925649
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,2,power_law_1.2,0.03175680041313171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,2,1,balanced,0.21251734097798666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,2,power_law_1.2,2.6237184524536135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,2,1,balanced,0.3813600142796834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,2,power_law_1.2,0.046489599347114566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,2,1,balanced,0.37989334265391034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,2,1,balanced,0.38580799102783203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,2,power_law_1.2,0.05980799794197082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,2,1,balanced,0.38742931683858234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,2,1,balanced,0.3930613199869792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,2,1,balanced,0.38674132029215497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,2,1,balanced,0.3910133441289266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,2,1,balanced,0.39424534638722736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,2,power_law_1.2,2.1887744903564452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,2,1,balanced,0.39983999729156494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,2,power_law_1.2,0.3218559980392456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,2,power_law_1.2,0.3507904052734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,2,power_law_1.2,2.147513580322266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,2,power_law_1.2,0.3729856014251709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,2,1,balanced,0.04896000027656555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,2,1,balanced,0.06798933446407318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,2,1,balanced,0.10583466291427612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,2,power_law_1.2,0.4017024040222168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,2,power_law_1.2,0.07418879866600037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,2,power_law_1.2,0.42902398109436035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,2,power_law_1.2,0.077183997631073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,2,power_law_1.2,0.5410048007965088
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,2,power_law_1.2,0.11548160314559937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,2,power_law_1.2,0.12155519723892212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,2,power_law_1.2,0.6195648193359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,2,power_law_1.2,0.12738560438156127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,2,power_law_1.2,0.13404159545898436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,2,power_law_1.2,0.8083711624145508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,2,power_law_1.2,0.13454079627990723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,2,1,balanced,0.17363733053207397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,2,power_law_1.2,0.14644479751586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,2,1,balanced,0.29792000850041706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,2,1,balanced,0.30085867643356323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,2,power_law_1.2,0.975443172454834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,2,power_law_1.2,0.13610880374908446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,2,1,balanced,0.3025760054588318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,2,1,balanced,0.3022773265838623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,2,power_law_1.2,0.14485119581222533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,2,1,balanced,0.3042400081952413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,2,1,balanced,0.30701865752538043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,2,power_law_1.2,1.3919232368469239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,2,1,balanced,0.40675731499989826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,2,1,balanced,0.5124693314234415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,2,1,balanced,0.5210613409678141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,2,power_law_1.2,3.045414352416992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,2,1,balanced,0.45926932493845624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,2,1,balanced,0.4803626537322998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,2,1,balanced,0.5057013432184855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,2,power_law_1.2,1.7723583221435546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,2,1,balanced,0.5942293405532837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,2,1,balanced,0.6393119891484579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,2,1,balanced,0.8369973500569662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,2,power_law_1.2,0.15460480451583863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,2,1,balanced,0.30691732962926227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,2,power_law_1.2,0.1512895941734314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,2,power_law_1.2,0.1572543978691101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,2,power_law_1.2,3.1916032791137696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,2,power_law_1.2,0.173363196849823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,2,power_law_1.2,0.18252160549163818
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,2,power_law_1.2,0.20485761165618896
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,2,power_law_1.2,0.24374399185180665
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,2,power_law_1.2,0.29027841091156004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,2,1,balanced,0.9158079624176025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,2,power_law_1.2,0.3463680028915405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,2,1,balanced,0.33770132064819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,2,1,balanced,0.34084800879160565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,2,1,balanced,1.417301336924235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,2,1,balanced,0.35258134206136066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,2,1,balanced,0.3688853184382121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,2,1,balanced,0.3819626569747925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,2,1,balanced,1.5783146222432454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,2,1,balanced,0.4007253249486287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,2,1,balanced,0.44839998086293537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,2,1,balanced,0.49770132700602215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,2,1,balanced,2.4261813163757324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,2,power_law_1.2,0.42031359672546387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,2,1,balanced,3.064549446105957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,2,power_law_1.2,2.5029632568359377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,2,1,balanced,0.045594667394955955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,2,1,balanced,0.060090666015942894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,2,1,balanced,0.08243200182914734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,2,1,balanced,0.7538560231526693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,2,1,balanced,0.1491200029850006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,2,1,balanced,0.2358026703198751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,2,1,balanced,0.24926400184631348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,2,1,balanced,0.2392586668332418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,2,1,balanced,0.24061334133148193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,2,1,balanced,0.24339733521143594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,2,1,balanced,0.24405866861343384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,2,1,balanced,0.24755734205245972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,2,1,balanced,0.25057599941889447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,2,1,balanced,0.255237340927124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,2,1,balanced,0.26360533634821576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,2,1,balanced,0.3643360137939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,2,1,balanced,0.3701440095901489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,2,1,balanced,0.3670986493428548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,2,1,balanced,0.3954879840215047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,2,1,balanced,0.42926931381225586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,2,1,balanced,0.8569227059682211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,2,1,balanced,0.6514079968134562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,2,1,balanced,0.7109546661376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,2,1,balanced,1.2130560080210369
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,2,power_law_1.2,0.5821184158325196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,2,1,balanced,1.0026826858520508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,2,1,balanced,1.57096529006958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,2,1,balanced,1.3110933303833008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,2,1,balanced,2.28983465830485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,2,1,balanced,1.909813404083252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,2,power_law_1.2,4.115161514282226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,2,1,balanced,3.0143038431803384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,2,1,balanced,4.436992009480794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,2,1,balanced,2.52509339650472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,2,power_law_1.2,4.40563850402832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,2,1,balanced,5.873008092244466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,2,1,balanced,3.6974827448527017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,2,1,balanced,4.920021375020345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,2,1,power_law_1.01,0.044563201069831845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,2,1,power_law_1.01,0.06980479955673217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,2,1,power_law_1.01,0.11721600294113159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,2,1,power_law_1.01,0.1648576021194458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,2,1,power_law_1.01,0.2143104076385498
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,2,1,balanced,0.02922666569550832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,2,1,balanced,0.034245334565639496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,2,1,balanced,0.053823997577031456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,2,1,balanced,0.07829333345095317
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,2,1,balanced,0.12550399700800577
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,2,1,balanced,0.12786133090655008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,2,1,balanced,0.1318719983100891
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,2,1,balanced,0.1338879962762197
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,2,1,balanced,0.136272003253301
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,2,1,balanced,0.13797866304715475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,2,1,balanced,0.14005333185195923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,2,1,balanced,0.1609493295351664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,2,1,balanced,0.16636266311009726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,2,1,balanced,0.17068799336751303
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,2,1,balanced,0.16902933518091837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,2,1,balanced,0.17327467600504556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,2,1,balanced,0.18203200896581015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,2,1,balanced,0.1971786618232727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,2,1,power_law_1.01,0.32140159606933594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,2,1,power_law_1.01,0.3306879997253418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,2,power_law_1.2,0.6907711982727051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,2,1,power_law_1.01,0.3468415975570679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,2,1,power_law_1.01,0.3575040102005005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,2,1,power_law_1.01,0.36501760482788087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,2,1,power_law_1.01,0.3832639932632446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,2,1,power_law_1.01,0.40163841247558596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,2,1,balanced,0.2068906625111898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,2,1,balanced,0.2790880004564921
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,2,1,balanced,0.25990400711695355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,2,1,balanced,0.5533013343811035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,2,1,balanced,0.3431520064671834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,2,1,balanced,0.4877279996871948
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,2,1,balanced,0.6253866751988729
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,2,1,balanced,0.8972799777984619
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,2,1,balanced,1.1779627005259197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,2,1,power_law_1.01,0.4128511905670166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,2,1,power_law_1.01,0.040652799606323245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,2,1,power_law_1.01,0.4444159984588623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,2,1,power_law_1.01,0.06538879871368408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,2,1,power_law_1.01,0.10515199899673462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,2,1,power_law_1.01,0.475705623626709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,2,1,power_law_1.01,0.14073599576950074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,2,1,power_law_1.01,0.5291840076446533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,2,1,power_law_1.01,0.17918720245361328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,2,1,power_law_1.01,0.47495040893554685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,2,1,power_law_1.01,0.2517119884490967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,2,1,power_law_1.01,0.26778240203857423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,2,1,power_law_1.01,0.5263552188873291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,2,1,power_law_1.01,0.2710400104522705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,2,1,power_law_1.01,0.5832960128784179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,2,1,power_law_1.01,0.29317119121551516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,2,1,power_law_1.01,0.655404806137085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,2,1,power_law_1.01,0.30963199138641356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,2,1,power_law_1.01,0.3461951971054077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,2,1,power_law_1.01,0.7516032218933105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,2,1,power_law_1.01,0.3297919988632202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,2,1,power_law_1.01,0.3463680028915405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,2,1,power_law_1.01,1.0080575942993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,2,1,power_law_1.01,0.36903679370880127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,2,1,power_law_1.01,0.38448638916015626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,2,1,power_law_1.01,0.40792322158813477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,2,1,power_law_1.01,0.4540031909942627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,2,1,power_law_1.01,0.5311679840087891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,2,1,power_law_1.01,0.6195136070251465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,2,1,power_law_1.01,0.7971327781677247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,2,power_law_1.2,3.301375961303711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,2,1,power_law_1.01,1.149120044708252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,2,1,power_law_1.01,0.9803071975708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,2,1,power_law_1.2,1.9476352691650392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,2,1,power_law_1.01,1.5240127563476562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,2,1,power_law_1.01,1.332582378387451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,2,1,power_law_1.01,1.8791807174682618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,2,1,power_law_1.01,1.6902784347534179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,2,1,power_law_1.01,0.0412992000579834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,2,power_law_1.2,6.310809707641601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,2,1,power_law_1.01,0.057036799192428586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,2,1,power_law_1.01,0.07953280210494995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,2,1,power_law_1.01,2.58538875579834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,2,1,power_law_1.01,2.4078847885131838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,2,1,power_law_1.01,0.11611520051956177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,2,1,power_law_1.01,0.15043840408325196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,2,1,power_law_1.01,0.21001598834991456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,2,1,power_law_1.01,3.126163291931152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,2,1,power_law_1.01,0.223801589012146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,2,1,power_law_1.01,0.2257920026779175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,2,1,power_law_1.01,0.23011200428009032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,2,1,power_law_1.01,0.2335871934890747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,2,1,power_law_1.01,0.2524800062179565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,2,1,power_law_1.01,0.266483211517334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,2,1,power_law_1.01,4.554361724853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,2,1,power_law_1.01,0.27927680015563966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,2,1,power_law_1.01,3.4010753631591797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,2,1,power_law_1.01,0.3133631944656372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,2,1,power_law_1.01,0.34511361122131345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,2,1,power_law_1.01,0.3838399887084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,2,1,power_law_1.01,0.3912832021713257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,2,1,power_law_1.01,0.45377278327941895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,2,1,power_law_1.01,5.99059829711914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,2,1,power_law_1.01,0.5312575817108154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,2,1,power_law_1.2,0.0455808013677597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,2,1,power_law_1.2,0.07126399874687195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,2,1,power_law_1.01,0.6814015865325928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,2,1,power_law_1.2,0.11413120031356812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,2,1,power_law_1.2,0.15381120443344115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,2,1,power_law_1.2,0.2038655996322632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,2,1,power_law_1.2,0.32497920989990237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,2,1,power_law_1.01,0.028217598795890808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,2,1,power_law_1.2,0.3360703945159912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,2,1,power_law_1.2,0.3554368019104004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,2,1,power_law_1.2,0.37180800437927247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,2,1,power_law_1.2,0.3664639949798584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,2,1,power_law_1.01,0.8268351554870605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,2,1,power_law_1.2,0.39140479564666747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,2,1,power_law_1.2,0.40247039794921874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,2,1,power_law_1.01,1.1349696159362792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,2,1,power_law_1.2,0.4174208164215088
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,2,1,power_law_1.01,0.03706879913806915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,2,1,power_law_1.01,0.05420799851417542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,2,1,power_law_1.2,0.44657278060913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,2,1,power_law_1.01,1.4164671897888184
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,2,1,power_law_1.01,0.06653439998626709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,2,1,power_law_1.2,0.4793344020843506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,2,1,power_law_1.01,0.08369280099868774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,2,1,power_law_1.2,0.5384191989898681
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,2,1,power_law_1.01,0.11397119760513305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,2,1,power_law_1.01,0.12023680210113526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,2,1,power_law_1.01,2.011142349243164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,2,1,power_law_1.2,0.4790207862854004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,2,1,power_law_1.01,0.1260032057762146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,2,1,power_law_1.2,0.534771203994751
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,2,1,power_law_1.01,0.13229440450668334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,2,1,power_law_1.01,0.13611520528793336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,2,1,power_law_1.2,0.5985919952392578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,2,1,power_law_1.01,0.144377601146698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,2,1,power_law_1.01,2.593753623962402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,2,1,power_law_1.2,0.6717376232147216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,2,1,power_law_1.01,0.15367679595947265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,2,1,power_law_1.01,0.16437760591506959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,2,1,power_law_1.2,0.7744832038879395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,2,power_law_1.2,0.9746047973632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,2,1,power_law_1.01,0.1845247983932495
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,2,1,power_law_1.01,0.1664896011352539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,2,1,power_law_1.2,1.0451199531555175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,2,1,power_law_1.01,3.7704639434814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,2,1,power_law_1.01,0.1756351947784424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,2,1,power_law_1.01,4.94532470703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,2,1,power_law_1.01,0.19853440523147584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,2,1,power_law_1.2,0.04708479940891266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,2,1,power_law_1.01,0.20200960636138915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,2,1,power_law_1.2,0.06445440053939819
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,2,1,power_law_1.01,0.23382399082183838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,2,1,power_law_1.2,0.10401279926300049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,2,1,power_law_1.2,1.1986047744750976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,2,1,power_law_1.2,0.12970880270004273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,2,1,power_law_1.2,0.15813759565353394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,2,1,power_law_1.2,0.2537472009658813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,2,1,power_law_1.2,0.26645760536193847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,2,1,power_law_1.2,0.28533759117126467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,2,1,power_law_1.2,0.29820799827575684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,2,1,power_law_1.01,0.24939520359039308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,2,1,power_law_1.2,0.31945600509643557
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,2,1,power_law_1.01,0.2902656078338623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,2,1,power_law_1.2,0.3487168073654175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,2,1,power_law_1.01,0.3235775947570801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,2,1,power_law_1.2,0.3391551971435547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,2,1,power_law_1.01,0.3910399913787842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,2,1,power_law_1.2,0.3517632007598877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,2,1,power_law_1.2,0.37426559925079345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,2,1,power_law_1.2,0.3928512096405029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,2,1,power_law_1.2,0.41797761917114257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,2,1,power_law_1.2,0.46245760917663575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,2,1,power_law_1.2,0.5442431926727295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,2,1,power_law_1.2,0.6324736118316651
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,2,1,power_law_1.01,0.5276927947998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,2,1,power_law_1.2,0.8112192153930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,2,1,power_law_1.01,0.670963191986084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,2,1,power_law_1.2,1.5400192260742187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,2,1,power_law_1.01,0.9467647552490235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,2,1,power_law_1.2,0.9877120018005371
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,2,1,power_law_1.01,1.3427136421203614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,2,1,power_law_1.2,0.0420415997505188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,2,1,power_law_1.2,0.057734400033950806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,2,1,power_law_1.2,0.07748479843139648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,2,1,power_law_1.2,0.10307199954986572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,2,1,power_law_1.2,1.3439999580383302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,2,1,power_law_1.2,0.1483199954032898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,2,1,power_law_1.2,0.20992639064788818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,2,1,power_law_1.2,0.2233344078063965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,2,1,power_law_1.2,0.22978560924530028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,2,1,power_law_1.2,0.23142399787902831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,2,1,power_law_1.2,0.23355519771575928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,2,1,power_law_1.2,0.2576767921447754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,2,1,power_law_1.2,0.2642751932144165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,2,1,power_law_1.2,0.2803839921951294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,2,1,power_law_1.2,0.31702399253845215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,2,1,power_law_1.2,1.8892288208007812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,2,1,power_law_1.2,0.3512192010879517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,2,1,power_law_1.2,0.39130239486694335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,2,1,power_law_1.2,0.39422080516815183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,2,1,power_law_1.2,1.7086847305297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,2,1,power_law_1.2,0.4705215930938721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,2,1,power_law_1.2,0.5430912017822266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,2,1,power_law_1.2,2.4165952682495115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,2,1,power_law_1.2,0.6953023910522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,2,1,power_law_1.2,0.8309056282043457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,2,1,power_law_1.2,1.1324159622192382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,2,1,power_law_1.2,3.1351871490478516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,2,1,power_law_1.2,1.4225791931152343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,2,power_law_1.2,4.871993637084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,2,1,power_law_1.2,2.016371154785156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,2,1,power_law_1.2,2.604198455810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,2,1,power_law_1.2,4.5703166961669925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,2,1,power_law_1.2,3.7776126861572266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,2,1,power_law_1.2,0.6252416133880615
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,2,power_law_1.2,1.6058176040649415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,2,balanced,0.10761599739392598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,2,balanced,0.16311466693878174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,2,balanced,0.3241013288497925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,2,balanced,0.5311839977900187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,2,balanced,0.995306650797526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,2,1,power_law_1.2,0.028326401114463808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,2,balanced,1.258677323659261
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,2,1,power_law_1.2,0.03808639943599701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,2,balanced,1.2630613644917805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,2,1,power_law_1.2,0.05279359817504883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,2,1,power_law_1.2,0.06228479743003845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,2,balanced,1.2615893681844075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,2,1,power_law_1.2,0.081740802526474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,2,balanced,1.2699466546376545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,2,1,power_law_1.2,0.1143231987953186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,2,1,power_law_1.2,0.1210752010345459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,2,balanced,1.3819786707560222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,2,power_law_1.2,8.291020965576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,2,1,power_law_1.2,0.12810239791870118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,2,balanced,1.389893372853597
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,2,1,power_law_1.2,0.13397760391235353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,2,balanced,0.09037333726882935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,2,balanced,0.1360106666882833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,2,balanced,0.20147200425465903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,2,1,power_law_1.2,0.13685120344161988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,2,balanced,0.3547626733779907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,2,1,power_law_1.2,0.14565759897232056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,2,balanced,0.6564000050226847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,2,1,power_law_1.2,0.15700479745864868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,2,balanced,0.8044053713480631
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,2,1,power_law_1.2,0.16607359647750855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,2,balanced,0.8088213602701823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,2,1,power_law_1.2,0.18703360557556153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,2,balanced,0.8107733726501465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,2,1,power_law_1.2,0.16626559495925902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,2,balanced,0.8149813016255697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,2,1,power_law_1.2,0.1783360004425049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,2,balanced,0.85044264793396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,2,1,power_law_1.2,0.2038719892501831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,2,balanced,0.8616853555043539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,2,balanced,1.3988587061564128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,2,1,power_law_1.2,0.2148672103881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,2,balanced,0.8743253548940023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,2,balanced,1.414789358774821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,2,balanced,0.8859946727752686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,2,balanced,0.9084906578063965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,2,power_law_1.2,6.314188766479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,2,balanced,1.4312373797098796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,2,balanced,0.9325599670410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,2,balanced,1.4513813654581706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,2,balanced,0.9673813184102377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,2,balanced,1.02237868309021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,2,balanced,1.1496373017628987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,2,balanced,1.274832010269165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,2,balanced,1.969530741373698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,2,1,power_law_1.2,6.010566329956054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,2,balanced,2.2088425954182944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,2,balanced,1.4755573272705078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,2,balanced,1.5130987167358398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,2,1,power_law_1.2,0.23816320896148682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,2,balanced,3.20523738861084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,2,balanced,1.595695972442627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,2,1,power_law_1.2,0.25518720149993895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,2,balanced,0.08499733606974284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,2,balanced,1.6896640459696453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,2,balanced,0.1402293344338735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,2,balanced,0.208186666170756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,2,balanced,0.35235734780629474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,2,1,power_law_1.2,0.29665279388427734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,2,balanced,0.6338293155034384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,2,balanced,4.178906758626302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,2,balanced,2.464394728342692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,2,balanced,0.7809546788533529
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,2,1,power_law_1.2,0.33063039779663084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,2,balanced,0.7886719703674316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,2,balanced,0.7930400371551514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,2,balanced,2.7361440658569336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,2,balanced,0.7995786666870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,2,balanced,0.8071253299713135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,2,balanced,5.697711944580078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,2,balanced,0.817530632019043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,2,balanced,3.8920586903889975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,2,balanced,0.8302453358968099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,2,balanced,0.8417973518371582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,2,1,power_law_1.2,2.5749568939208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,2,balanced,4.706495920817058
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,2,1,power_law_1.2,0.39946880340576174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,2,balanced,7.659498850504558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,2,balanced,0.8643999894460043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,2,balanced,0.8855413595835367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,2,balanced,6.80351448059082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,2,balanced,0.9080213705698649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,2,balanced,0.9465013345082601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,2,balanced,1.0212106704711914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,2,balanced,1.0976320107777913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,2,balanced,11.234837849934896
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,2,1,power_law_1.2,0.5411839962005616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,2,1,power_law_1.2,4.951366424560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,2,balanced,10.169008255004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,2,balanced,1.6597119967142742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,2,balanced,1.8099145889282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,2,balanced,2.5762292544047036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,2,balanced,16.63435236612956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,2,balanced,3.3421014149983725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,2,power_law_1.01,0.14901119470596313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,2,balanced,15.585306803385416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,2,power_law_1.01,0.316211199760437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,2,power_law_1.01,0.3808896064758301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,2,balanced,4.44432004292806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,2,balanced,0.043653334180514015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,2,balanced,0.0621973325808843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,2,balanced,0.09402133027712505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,2,balanced,0.15100266536076865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,2,balanced,0.26732800404230755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,2,balanced,0.32603200276692706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,2,balanced,5.982048034667969
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,2,balanced,0.3293546636899312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,2,balanced,0.33485865592956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,2,power_law_1.01,0.5770175933837891
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,2,balanced,0.33666133880615234
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,2,balanced,0.340554674466451
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,2,balanced,0.34778666496276855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,2,power_law_1.01,0.6204991817474366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,2,balanced,0.3545600175857544
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,2,balanced,0.34671465555826825
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,2,balanced,0.35687466462453205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,2,power_law_1.01,0.9713151931762696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,2,balanced,0.36512001355489093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,2,balanced,0.35948801040649414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,2,balanced,0.37062398592631024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,2,power_law_1.01,1.0942079544067382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,2,balanced,0.40969598293304443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,2,balanced,0.400490681330363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,2,balanced,8.59599494934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,2,balanced,0.6924959818522135
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,2,balanced,0.5052746534347534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,2,balanced,21.375775655110676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,2,balanced,1.2176746527353923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,2,1,power_law_1.2,0.6818751811981201
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,2,balanced,0.8220907052357992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,2,power_law_1.01,0.12855679988861085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,2,balanced,1.180341323216756
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,2,balanced,1.5248533884684246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,2,power_law_1.01,1.1068736076354981
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,2,balanced,1.714570681254069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,2,power_law_1.01,1.1417728424072267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,2,balanced,11.187204996744791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,2,balanced,2.710426648457845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,2,power_law_1.01,0.19468159675598146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,2,power_law_1.01,0.25452160835266113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,2,power_law_1.01,0.34787840843200685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,2,power_law_1.01,0.45159039497375486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,2,power_law_1.01,0.6512896060943604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,2,power_law_1.01,0.13861759901046752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,2,power_law_1.01,1.2841279983520508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,2,power_law_1.01,0.19771519899368287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,2,power_law_1.01,0.2552704095840454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,2,power_law_1.01,1.3209600448608398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,2,power_law_1.01,0.3334208011627197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,2,power_law_1.01,0.42972798347473146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,2,power_law_1.01,1.3723648071289063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,2,power_law_1.01,0.6841087818145752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,2,power_law_1.01,1.3750528335571288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,2,power_law_1.01,0.6864704132080078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,2,power_law_1.01,1.3628031730651855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,2,power_law_1.01,0.7267007827758789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,2,power_law_1.01,0.7212031841278076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,2,power_law_1.01,1.4608511924743652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,2,power_law_1.01,0.7454783916473389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,2,power_law_1.01,1.494713592529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,2,power_law_1.01,0.7448383808135987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,2,power_law_1.01,0.7096191883087158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,2,power_law_1.01,0.7702720165252686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,2,power_law_1.01,1.5879743576049805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,2,power_law_1.01,0.6998847961425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,2,power_law_1.01,0.06119040250778198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,2,power_law_1.01,0.798137617111206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,2,power_law_1.01,0.7435520172119141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,2,power_law_1.01,1.8166400909423828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,2,power_law_1.01,0.8303104400634765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,2,power_law_1.01,0.7805183887481689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,2,power_law_1.01,0.8728256225585938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,2,power_law_1.01,0.8256447792053223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,2,power_law_1.01,1.9944000244140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,2,power_law_1.01,0.9052288055419921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,2,power_law_1.01,0.9910976409912109
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,2,power_law_1.01,0.08303359746932984
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,2,power_law_1.01,0.11651840209960937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,2,power_law_1.01,1.1516096115112304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,2,power_law_1.01,0.15720319747924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,2,power_law_1.01,0.18883199691772462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,2,power_law_1.01,1.3378687858581544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,2,power_law_1.01,0.8547391891479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,2,power_law_1.01,0.865721607208252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,2,power_law_1.01,1.7023616790771485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,2,power_law_1.01,2.5128320693969726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,2,power_law_1.01,0.8902015686035156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,2,power_law_1.2,6.368582534790039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,2,power_law_1.01,1.9865087509155273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,2,power_law_1.2,0.1504320025444031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,2,power_law_1.01,2.875961685180664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,2,power_law_1.2,0.2558336019515991
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,2,power_law_1.01,0.2607935905456543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,2,power_law_1.01,0.2857088088989258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,2,power_law_1.2,0.3788288116455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,2,power_law_1.01,2.7233919143676757
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,2,power_law_1.01,0.2982719898223877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,2,power_law_1.2,0.5013696193695069
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,2,power_law_1.01,0.3050944089889526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,2,power_law_1.2,0.6685952186584473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,2,power_law_1.01,3.957689666748047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,2,power_law_1.01,0.9530431747436523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,2,power_law_1.2,1.0076416015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,2,power_law_1.01,3.553094482421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,2,power_law_1.2,1.036620807647705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,2,power_law_1.2,1.1139455795288087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,2,power_law_1.01,5.188755035400391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,2,power_law_1.01,0.32084479331970217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,2,power_law_1.2,1.1388607978820802
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,2,power_law_1.01,0.341644811630249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,2,power_law_1.01,4.727590560913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,2,power_law_1.01,0.36060800552368166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,2,power_law_1.2,1.2790783882141112
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,2,power_law_1.01,0.3236991882324219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,2,power_law_1.01,0.36599040031433105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,2,power_law_1.2,1.3499584197998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,2,power_law_1.01,0.3767935991287231
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,2,power_law_1.01,0.3613312005996704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,2,power_law_1.2,1.3796671867370605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,2,power_law_1.01,6.918547058105469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,2,power_law_1.01,0.39467520713806153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,2,power_law_1.2,1.3758336067199708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,2,power_law_1.01,6.4770751953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,2,power_law_1.01,1.0018303871154786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,2,power_law_1.2,1.4262720108032227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,2,power_law_1.01,1.10830078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,2,power_law_1.01,1.3301119804382324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,2,power_law_1.2,1.5178560256958007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,2,power_law_1.01,1.549407958984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,2,power_law_1.01,9.816070556640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,2,power_law_1.01,2.057734489440918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,2,power_law_1.01,9.901913452148438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,2,power_law_1.01,2.4458303451538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,2,power_law_1.2,1.542188835144043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,2,power_law_1.2,1.6251712799072267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,2,power_law_1.2,8.1574462890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,2,power_law_1.01,3.4552639007568358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,2,power_law_1.2,0.12400000095367432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,2,power_law_1.2,1.8495807647705078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,2,power_law_1.2,0.17429120540618898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,2,power_law_1.2,0.24411520957946778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,2,power_law_1.2,0.3471935987472534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,2,power_law_1.01,4.5642047882080075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,2,power_law_1.2,0.39610240459442136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,2,power_law_1.2,0.6331456184387207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,2,power_law_1.01,11.865644836425782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,2,power_law_1.2,0.7002624034881592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,2,power_law_1.2,0.6904255867004394
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,2,power_law_1.01,0.41619200706481935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,2,power_law_1.2,0.7792768001556396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,2,power_law_1.2,2.095052719116211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,2,power_law_1.01,0.4510079860687256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,2,power_law_1.01,6.137766265869141
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,2,power_law_1.01,0.5948800086975098
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,2,power_law_1.01,0.6270847797393799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,2,power_law_1.2,2.5717184066772463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,2,power_law_1.2,0.13700480461120607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,2,power_law_1.01,15.129689025878907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,2,power_law_1.2,0.15198080539703368
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,2,power_law_1.01,0.7349120140075683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,2,power_law_1.2,0.2361664056777954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,2,power_law_1.01,0.8699520111083985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,2,power_law_1.2,0.3429951906204224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,2,power_law_1.2,3.007583999633789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,2,power_law_1.2,0.785868787765503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,2,power_law_1.2,0.368287992477417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,2,power_law_1.01,1.1093695640563965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,2,power_law_1.2,0.8430399894714355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,2,power_law_1.2,0.6497087955474854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,2,power_law_1.01,8.390425872802734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,2,power_law_1.01,1.4575872421264648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,2,power_law_1.2,0.8671039581298828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,2,power_law_1.2,0.6871232032775879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,2,power_law_1.2,4.289689636230468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,2,power_law_1.2,0.8654911994934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,2,power_law_1.01,1.9805376052856445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,2,power_law_1.2,0.9300736427307129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,2,power_law_1.2,0.9732159614562989
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,2,1,power_law_1.2,0.9521727561950684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,2,power_law_1.2,1.0419008255004882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,2,power_law_1.01,3.1832895278930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,2,power_law_1.2,5.271539306640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,2,power_law_1.2,0.7068480014801025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,2,power_law_1.2,1.1846976280212402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,2,power_law_1.2,0.7368896007537842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,2,power_law_1.2,1.3584383964538573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,2,power_law_1.2,0.7554304122924804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,2,power_law_1.01,18.403872680664062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,2,power_law_1.01,12.109478759765626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,2,power_law_1.2,1.604921531677246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,2,1,balanced,0.09898133079210918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,2,1,balanced,0.15221866965293884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,2,1,balanced,0.3166133364041646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,2,1,balanced,0.5171786546707153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,2,power_law_1.2,2.080793571472168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,2,power_law_1.2,7.742566680908203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,2,power_law_1.2,2.488921546936035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,2,power_law_1.2,0.7833663940429687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,2,power_law_1.2,0.7946303844451904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,2,1,balanced,1.0020480155944824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,2,power_law_1.2,0.8107968330383301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,2,power_law_1.2,3.516543960571289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,2,1,balanced,1.2349173227945964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,2,power_law_1.2,0.8548031806945801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,2,1,balanced,1.2379199663798015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,2,1,balanced,1.2413866519927979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,2,power_law_1.2,0.9016127586364746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,2,1,balanced,1.2449226379394531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,2,power_law_1.2,0.9410559654235839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,2,1,balanced,1.2516000270843506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,2,power_law_1.2,4.697241592407226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,2,power_law_1.2,0.06124160289764404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,2,1,balanced,1.2612213293711345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,2,power_law_1.2,1.0319487571716308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,2,power_law_1.2,10.217510223388672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,2,power_law_1.2,0.07200000286102295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,2,1,balanced,1.2652479807535808
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,2,power_law_1.2,0.10574079751968384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,2,power_law_1.2,1.200160026550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,2,power_law_1.2,0.13971840143203734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,2,power_law_1.2,0.17621760368347167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,2,power_law_1.2,1.3565888404846191
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,2,power_law_1.2,0.27139840126037595
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,2,power_law_1.2,0.28691198825836184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,2,power_law_1.2,6.727302551269531
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,2,power_law_1.2,0.294598388671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,2,power_law_1.01,16.677510070800782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,2,power_law_1.2,1.7636800765991212
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,2,power_law_1.2,0.30817279815673826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,2,1,balanced,0.08455999692281087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,2,1,balanced,0.11638933420181274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,2,power_law_1.2,0.3203775882720947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,2,1,balanced,1.3806773821512859
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,2,power_law_1.2,0.3524928092956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,2,power_law_1.2,2.065113639831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,2,1,balanced,1.3928319613138835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,2,power_law_1.2,0.3692608118057251
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,2,power_law_1.2,0.34225919246673586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,2,power_law_1.2,0.37017600536346434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,2,power_law_1.2,2.806752014160156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,2,power_law_1.2,0.3930047988891602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,2,1,balanced,0.19171200195948282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,2,1,balanced,0.3547626733779907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,2,1,power_law_1.2,3.2723648071289064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,2,1,balanced,0.6668000221252441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,2,1,balanced,0.8188373247782389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,2,power_law_1.2,3.662150573730469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,2,1,balanced,0.8213866551717123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,2,1,balanced,1.404197374979655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,2,power_law_1.2,8.623725128173827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,2,1,balanced,0.8232853412628174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,2,1,balanced,1.4211039543151855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,2,1,balanced,0.8247360388437907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,2,1,balanced,0.8280746936798096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,2,1,balanced,1.4569973945617676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,2,power_law_1.2,0.3740992069244385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,2,1,balanced,0.8326079845428467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,2,1,balanced,0.08177599807580312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,2,1,balanced,0.8399519920349121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,2,power_law_1.2,0.40430078506469724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,2,1,balanced,0.8487520217895508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,2,power_law_1.2,0.4338111877441406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,2,power_law_1.2,15.91808624267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,2,1,balanced,0.8665706316630045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,2,power_law_1.2,0.4691648006439209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,2,1,balanced,0.8861066500345866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,2,1,balanced,0.9127253691355387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,2,power_law_1.2,0.6549183845520019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,2,1,balanced,0.955952008565267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,2,power_law_1.2,0.6419583797454834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,2,1,balanced,1.0615999698638916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,2,1,balanced,0.1317759950955709
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,2,power_law_1.2,0.7474112033843994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,2,1,balanced,1.1670986811319988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,2,power_law_1.2,5.120633697509765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,2,power_law_1.2,0.9178496360778808
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,2,power_law_1.2,1.208140754699707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,2,1,balanced,1.5386133193969727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,2,1,balanced,0.21002666155497232
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,2,power_law_1.2,1.5488832473754883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,2,1,balanced,1.6211999257405598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,2,1,balanced,1.8152586619059246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,2,1,balanced,2.290341377258301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,2,1,balanced,2.008000055948893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,2,1,balanced,2.576325257619222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,2,1,balanced,2.8998400370279946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,2,1,balanced,0.3628426790237427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,2,1,balanced,0.6513599952061971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,2,power_law_1.2,6.39148178100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,2,1,balanced,0.8003040154774984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,2,1,balanced,3.8008267084757485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,2,1,balanced,0.8055253028869629
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,2,power_law_1.2,2.032032012939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,2,1,balanced,0.8106239636739095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,2,1,balanced,0.8166986306508383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,2,1,balanced,0.8204053243001302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,2,1,balanced,0.8306506474812826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,2,1,balanced,5.115018526713054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,2,1,balanced,0.8398346900939941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,2,1,balanced,0.8467893600463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,2,1,balanced,3.6080214182535806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,2,1,balanced,0.8638827006022135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,2,1,balanced,0.8821653525034586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,2,1,balanced,0.8972640037536621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,2,1,balanced,0.9275519847869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,2,1,balanced,6.915370941162109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,2,1,balanced,0.9801920255025228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,2,1,balanced,1.0399733384450276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,2,1,balanced,1.6188586552937825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,2,1,balanced,1.7284266153971355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,2,1,balanced,4.331130663553874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,2,power_law_1.2,21.229190063476562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,2,1,balanced,2.477824052174886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,2,1,balanced,10.045370737711588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,2,power_law_1.2,13.283750915527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,2,1,balanced,3.21942933400472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,2,1,balanced,5.6594664255778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,2,1,balanced,4.193365414937337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,2,1,balanced,13.192949930826822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,2,1,balanced,7.885141372680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,2,1,power_law_1.01,0.091430401802063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,2,1,power_law_1.01,0.1476096034049988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,2,1,balanced,5.674645105997722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,2,1,power_law_1.01,0.31059200763702394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,2,1,power_law_1.01,0.46742401123046873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,2,1,power_law_1.01,0.6092607975006104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,2,power_law_1.2,3.4916927337646486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,2,1,balanced,0.0440586656332016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,2,1,balanced,0.06384533147017162
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,2,1,balanced,0.09134399890899658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,2,1,power_law_1.01,0.9857151985168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,2,1,balanced,0.1495039959748586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,2,1,balanced,0.26531734069188434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,2,1,power_law_1.01,0.0828927993774414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,2,1,balanced,0.3255573312441508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,2,1,balanced,0.3305973410606384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,2,1,power_law_1.01,1.0534144401550294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,2,1,balanced,8.101386388142904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,2,1,balanced,11.810426076253256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,2,1,power_law_1.01,1.051097583770752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,2,1,power_law_1.01,1.1124671936035155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,2,1,power_law_1.01,1.1078656196594239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,2,1,power_law_1.01,0.11354240179061889
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,2,1,balanced,0.3369973500569661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,2,1,power_law_1.01,1.1910976409912108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,2,1,power_law_1.01,0.1913472056388855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,2,1,power_law_1.01,0.2954751968383789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,2,1,balanced,10.496709187825521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,2,1,power_law_1.01,0.4164415836334229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,2,1,power_law_1.01,0.6382912158966064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,2,1,power_law_1.01,0.681440019607544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,2,1,power_law_1.01,0.7096704006195068
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,2,1,balanced,0.3387519915898641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,2,1,balanced,16.98859151204427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,2,1,balanced,0.34067734082539874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,2,1,balanced,0.34618135293324787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,2,1,power_law_1.01,1.2872575759887694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,2,1,power_law_1.01,0.0789247989654541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,2,1,balanced,0.35171735286712646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,2,1,balanced,0.3924533526102702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,2,1,power_law_1.01,0.12844159603118896
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,2,1,balanced,0.4046773513158162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,2,1,power_law_1.01,1.3065919876098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,2,1,balanced,0.41748801867167157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,2,1,power_law_1.01,0.2083967924118042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,2,1,balanced,0.40614934762318927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,2,1,power_law_1.01,0.30230400562286375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,2,1,power_law_1.01,1.3592960357666015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,2,1,power_law_1.01,0.41753602027893066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,2,1,power_law_1.01,0.6480639934539795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,2,power_law_1.2,9.910163116455077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,2,1,power_law_1.01,0.7379007816314698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,2,1,power_law_1.01,0.6843904018402099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,2,1,power_law_1.01,0.766918420791626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,2,1,power_law_1.01,0.7264319896697998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,2,1,power_law_1.01,0.8432127952575683
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,2,1,balanced,0.420415997505188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,2,1,power_law_1.01,0.7395904064178467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,2,1,balanced,0.4625866810480754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,2,1,power_law_1.01,0.9085375785827636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,2,1,balanced,0.4789973497390747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,2,1,power_law_1.01,0.7468416213989257
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,2,1,balanced,0.7191946506500244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,2,1,power_law_1.01,0.8071167945861817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,2,1,power_law_1.01,1.3632448196411133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,2,1,balanced,0.5987253189086914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,2,1,power_law_1.01,0.8545599937438965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,2,1,balanced,1.6320053736368816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,2,1,power_law_1.01,1.4365887641906738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,2,1,power_law_1.01,0.8872320175170898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,2,1,balanced,1.096981366475423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,2,power_law_1.2,15.949784851074218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,2,1,power_law_1.01,1.5218303680419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,2,1,power_law_1.01,0.9313599586486816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,2,1,balanced,1.5824106534322102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,2,1,power_law_1.2,0.09117439985275269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,2,1,power_law_1.2,0.1480831980705261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,2,1,power_law_1.01,1.0223039627075194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,2,1,power_law_1.01,0.04069119989871979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,2,1,power_law_1.01,1.6731903076171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,2,1,power_law_1.01,0.7741695880889893
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,2,1,power_law_1.01,0.060262399911880496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,2,1,power_law_1.01,1.218508815765381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,2,1,power_law_1.01,0.8071616172790528
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,2,1,power_law_1.01,0.08812159895896912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,2,1,power_law_1.01,1.8523839950561523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,2,1,power_law_1.01,0.12594560384750367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,2,1,power_law_1.01,0.7989952087402343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,2,1,power_law_1.01,1.39552640914917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,2,1,power_law_1.01,0.17176320552825927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,2,1,power_law_1.01,0.8155648231506347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,2,1,power_law_1.01,0.25772800445556643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,2,1,power_law_1.01,2.2833471298217773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,2,1,power_law_1.01,1.7905792236328124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,2,1,power_law_1.01,0.2809279918670654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,2,1,power_law_1.01,0.8852992057800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,2,1,power_law_1.2,0.29459199905395506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,2,1,balanced,2.0707146326700845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,2,1,power_law_1.01,0.2991872072219849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,2,1,power_law_1.01,0.9165056228637696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,2,1,power_law_1.2,0.4109951972961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,2,1,power_law_1.01,0.301689600944519
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,2,1,balanced,1.8339625994364421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,2,1,power_law_1.01,2.6416704177856447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,2,1,power_law_1.01,0.3129215955734253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,2,1,power_law_1.2,0.5685376167297364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,2,1,power_law_1.01,2.177292823791504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,2,1,power_law_1.01,0.9705408096313477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,2,1,power_law_1.01,0.333024001121521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,2,1,power_law_1.2,0.9422335624694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,2,1,power_law_1.01,0.3457855939865112
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,2,1,balanced,2.785242716471354
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,2,1,power_law_1.01,0.3686975955963135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,2,1,power_law_1.2,1.0059264183044434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,2,1,power_law_1.01,2.978227233886719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,2,1,power_law_1.01,0.41711997985839844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,2,1,power_law_1.2,0.08324480056762695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,2,1,power_law_1.2,1.0665087699890137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,2,1,power_law_1.01,0.4679296016693115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,2,1,power_law_1.2,0.11519999504089355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,2,1,power_law_1.2,0.19213440418243408
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,2,1,power_law_1.01,0.39860479831695556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,2,1,power_law_1.2,1.1193408012390136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,2,1,power_law_1.2,0.2772608041763306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,2,1,power_law_1.01,0.4549056053161621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,2,1,power_law_1.2,1.1820672035217286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,2,1,power_law_1.01,3.771468734741211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,2,1,power_law_1.01,1.1165311813354493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,2,1,power_law_1.01,0.4488959789276123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,2,1,power_law_1.01,0.5188608169555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,2,1,power_law_1.2,1.2596351623535156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,2,1,power_law_1.2,1.4611136436462402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,2,1,power_law_1.01,3.5562686920166016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,2,1,power_law_1.01,0.540780782699585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,2,1,power_law_1.2,1.3334400177001953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,2,1,power_law_1.01,0.6215487957000733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,2,1,power_law_1.2,1.329843235015869
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,2,1,power_law_1.01,0.7028223991394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,2,power_law_1.2,12.34436492919922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,2,1,power_law_1.01,5.3750465393066404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,2,1,power_law_1.2,0.3774336099624634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,2,1,power_law_1.01,4.429151916503907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,2,1,power_law_1.2,1.3563520431518554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,2,1,power_law_1.2,0.6409535884857178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,2,1,power_law_1.01,1.2587648391723634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,2,1,power_law_1.2,0.6785471916198731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,2,1,power_law_1.2,1.3998399734497071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,2,1,power_law_1.2,0.7347839832305908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,2,1,power_law_1.01,1.5745216369628907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,2,1,power_law_1.2,1.464192008972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,2,1,power_law_1.2,0.7452352046966553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,2,1,power_law_1.2,0.07858560085296631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,2,1,power_law_1.2,0.808351993560791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,2,1,power_law_1.01,1.876473617553711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,2,1,power_law_1.01,7.016108703613281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,2,1,power_law_1.01,0.8402688026428222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,2,1,power_law_1.01,6.473248291015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,2,1,power_law_1.2,0.886963176727295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,2,1,power_law_1.01,1.1274751663208007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,2,1,power_law_1.2,0.9626367568969727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,2,1,power_law_1.01,2.512563133239746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,2,1,power_law_1.01,1.3900799751281738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,2,1,power_law_1.2,0.8285183906555176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,2,1,power_law_1.2,0.8828991889953614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,2,1,power_law_1.2,0.1277567982673645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,2,1,power_law_1.2,1.561292839050293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,2,1,power_law_1.01,3.1460544586181642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,2,1,power_law_1.2,0.206329607963562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,2,1,power_law_1.01,1.9176128387451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,2,1,power_law_1.2,0.9042048454284668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,2,1,power_law_1.2,0.28793599605560305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,2,1,power_law_1.2,1.7195327758789063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,2,1,power_law_1.01,8.259225463867187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,2,1,power_law_1.2,0.9652416229248046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,2,1,power_law_1.2,0.3811455965042114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,2,1,power_law_1.01,10.23702392578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,2,1,power_law_1.2,1.8959680557250977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,2,1,power_law_1.2,1.0475199699401856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,2,1,power_law_1.01,4.3917182922363285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,2,1,power_law_1.2,1.226841640472412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,2,1,power_law_1.2,2.3178815841674805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,2,1,power_law_1.2,1.439743995666504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,2,1,power_law_1.2,2.6829248428344727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,2,1,power_law_1.2,0.6384895801544189
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,2,1,power_law_1.01,2.7739967346191405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,2,1,power_law_1.2,1.8250688552856444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,2,1,power_law_1.01,5.650732803344726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,2,balanced,0.10860266288121541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,2,balanced,0.16448533535003662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,2,balanced,0.3065386613210042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,2,1,power_law_1.2,2.2087615966796874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,2,balanced,0.5380426645278931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,2,1,power_law_1.2,3.594348907470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,2,1,power_law_1.01,12.296832275390624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,2,balanced,0.9592426617940267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,2,balanced,1.8330186208089192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,2,1,power_law_1.2,3.002016067504883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,2,1,power_law_1.01,13.510041809082031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,2,1,power_law_1.2,0.6843200206756592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,2,balanced,2.7333812713623047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,2,1,power_law_1.2,0.7126272201538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,2,1,power_law_1.2,4.6144256591796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,2,1,power_law_1.01,8.129043579101562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,2,1,power_law_1.2,0.04046719968318939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,2,1,power_law_1.2,0.7351679801940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,2,1,power_law_1.2,0.0607807993888855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,2,1,power_law_1.2,3.8043968200683596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,2,1,power_law_1.2,0.7417600154876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,2,1,power_law_1.2,0.08849920034408569
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,2,1,power_law_1.2,0.11831680536270142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,2,1,power_law_1.2,0.7588160037994385
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,2,1,power_law_1.2,0.1616320013999939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,2,1,power_law_1.2,0.8149760246276856
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,2,1,power_law_1.2,0.26296319961547854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,2,1,power_law_1.2,6.586656188964843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,2,1,power_law_1.2,0.27820799350738523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,2,balanced,0.09507733583450317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,2,balanced,0.12921599547068277
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,2,1,power_law_1.2,0.28764159679412843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,2,balanced,0.20110932985941568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,2,balanced,0.3322133421897888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,2,1,power_law_1.2,0.30506880283355714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,2,balanced,0.6170080105463663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,2,1,power_law_1.2,0.31781759262084963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,2,balanced,1.1868267059326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,2,1,power_law_1.2,0.341977596282959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,2,1,power_law_1.01,15.827558898925782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,2,1,power_law_1.2,0.3609663963317871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,2,1,power_law_1.01,10.588499450683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,2,1,power_law_1.2,5.412966537475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,2,1,power_law_1.2,0.8296832084655762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,2,1,power_law_1.2,0.8430912017822265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,2,balanced,2.7523199717203775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,2,1,power_law_1.2,8.53939208984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,2,1,power_law_1.2,0.8885312080383301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,2,balanced,2.743781407674154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,2,1,power_law_1.2,0.91146240234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,2,balanced,1.7592320442199707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,2,1,power_law_1.2,0.38629119396209716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,2,balanced,2.7527573903401694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,2,balanced,1.7623039881388347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,2,1,power_law_1.2,0.42427520751953124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,2,balanced,1.7651039759318035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,2,1,power_law_1.2,0.48378877639770507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,2,balanced,2.7660373051961265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,2,1,power_law_1.2,0.4048128128051758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,2,balanced,1.7705386479695637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,2,1,power_law_1.2,0.4690688133239746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,2,balanced,2.786095937093099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,2,balanced,1.8197867075602214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,2,1,power_law_1.2,7.038604736328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,2,1,power_law_1.2,0.4796288013458252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,2,1,power_law_1.2,1.0035136222839356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,2,balanced,1.8064212799072266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,2,1,power_law_1.2,0.5396480083465576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,2,balanced,2.8099308013916016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,2,1,power_law_1.2,0.55829758644104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,2,balanced,1.8047733306884766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,2,1,power_law_1.2,0.6435071945190429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,2,balanced,1.9835999806722004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,2,balanced,0.0886346697807312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,2,1,power_law_1.2,0.7244991779327392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,2,balanced,0.11366400122642517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,2,balanced,0.18680532773335776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,2,balanced,2.0169386863708496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,2,balanced,0.33033066987991333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,2,1,power_law_1.2,12.0255615234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,2,balanced,0.5963893334070841
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,2,1,power_law_1.2,0.8574784278869629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,2,balanced,2.0463786125183105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,2,balanced,1.148645321528117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,2,1,power_law_1.2,1.1272383689880372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,2,balanced,1.7068479855855305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,2,balanced,2.093461354573568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,2,balanced,3.099482536315918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,2,balanced,0.036474667489528656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,2,balanced,1.711626688639323
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,2,balanced,0.05818133552869161
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,2,balanced,0.08787199854850769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,2,balanced,2.199183940887451
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,2,balanced,0.1439253290494283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,2,balanced,0.2551893393198649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,2,balanced,3.1001386642456055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,2,balanced,0.4724693298339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,2,balanced,0.6829919815063477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,2,balanced,2.32858673731486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,2,balanced,0.6878399848937988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,2,balanced,0.6901493072509766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,2,balanced,3.1197706858317056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,2,balanced,2.560469309488932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,2,1,power_law_1.2,10.268985748291016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,2,1,power_law_1.2,1.2926976203918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,2,1,power_law_1.2,1.1230463981628418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,2,balanced,3.17796262105306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,2,balanced,2.8008639017740884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,2,balanced,1.7204052607218425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,2,1,power_law_1.2,1.5888128280639648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,2,balanced,1.7244319915771484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,2,balanced,3.2653706868489585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,2,balanced,3.26419734954834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,2,balanced,1.7403039932250977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,2,balanced,0.6933600107828776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,2,balanced,0.7009226481119791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,2,balanced,3.3398186365763345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,2,balanced,1.7546933492024739
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,2,balanced,0.7094559669494629
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,2,balanced,0.7179733117421468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,2,balanced,4.81386121114095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,2,balanced,1.7658294041951497
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,2,balanced,0.7313013076782227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,2,balanced,0.7462133566538492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,2,balanced,3.5149173736572266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,2,balanced,1.789141337076823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,2,balanced,0.7610186735788981
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,2,balanced,0.7511733373006185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,2,balanced,1.8154080708821614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,2,balanced,0.7879573504130045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,2,balanced,0.7894186973571777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,2,balanced,3.738346735636393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,2,balanced,1.838175932566325
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,2,balanced,0.8371573289235433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,2,balanced,0.8647572994232178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,2,balanced,1.8826452891031902
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,2,balanced,0.9231253465016683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,2,balanced,4.1119734446207685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,2,balanced,1.294864018758138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,2,balanced,1.0963466962178547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,2,1,power_law_1.2,1.9023359298706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,2,balanced,5.807680130004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,2,balanced,2.2247840563456216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,2,1,power_law_1.2,2.525312042236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,2,balanced,1.8017279307047527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,2,balanced,1.9642240206400554
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,2,balanced,3.074303944905599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,2,balanced,2.044282595316569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,2,balanced,7.877898534138997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,2,balanced,2.188602606455485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,2,power_law_1.01,0.15278719663619994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,2,balanced,6.132506688435872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,2,power_law_1.01,0.29786880016326905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,2,balanced,2.335477352142334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,2,1,power_law_1.2,1.3980480194091798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,2,1,power_law_1.2,13.485165405273438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,2,balanced,2.7503039042154946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,2,1,power_law_1.2,3.158963203430176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,2,1,power_law_1.2,1.9357311248779296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,2,balanced,6.609333038330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,2,power_law_1.01,0.11681280136108399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,2,balanced,10.943584442138672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,2,power_law_1.01,0.1981376051902771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,2,balanced,3.8430614471435547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,2,power_law_1.01,0.5161983966827393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,2,power_law_1.01,0.3281152009963989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,2,power_law_1.01,0.6605696201324462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,2,power_law_1.01,0.4271999835968018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,2,balanced,4.525199890136719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,2,power_law_1.01,0.6528192043304444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,2,power_law_1.01,1.044115161895752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,2,balanced,9.461722691853842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,2,power_law_1.01,0.8622464179992676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,2,power_law_1.01,1.3772735595703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,2,power_law_1.01,1.0662528038024903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,2,balanced,6.056954701741536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,2,power_law_1.01,1.5256319999694825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,2,power_law_1.01,1.1705535888671874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,2,balanced,15.767573038736979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,2,power_law_1.01,1.9577407836914062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,2,power_law_1.01,1.2372159957885742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,2,1,power_law_1.2,4.4049217224121096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,2,power_law_1.01,2.04015998840332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,2,balanced,8.325018564860025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,2,balanced,14.295237223307291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,2,power_law_1.01,2.0156864166259765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,2,1,power_law_1.2,5.635583877563477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,2,power_law_1.01,2.1998336791992186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,2,power_law_1.01,1.3366335868835448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,2,power_law_1.01,1.3712639808654785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,2,power_law_1.01,2.146291160583496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,2,balanced,11.578725179036459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,2,power_law_1.01,1.4500160217285156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,2,power_law_1.01,2.3879743576049806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,2,power_law_1.01,0.10859520435333252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,2,power_law_1.01,1.5361599922180176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,2,power_law_1.01,0.18175359964370727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,2,1,power_law_1.2,8.113779449462891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,2,power_law_1.01,2.5278400421142577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,2,power_law_1.01,0.3229759931564331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,2,power_law_1.01,1.6843135833740235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,2,power_law_1.01,0.4016831874847412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,2,power_law_1.01,1.7079999923706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,2,power_law_1.01,2.6805248260498047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,2,power_law_1.01,0.6654272079467773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,2,balanced,23.631940205891926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,2,power_law_1.01,0.8254783630371094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,2,power_law_1.01,1.7889087677001954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,2,power_law_1.01,2.6776256561279297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,2,power_law_1.2,0.15386879444122314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,2,power_law_1.01,1.0515328407287599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,2,power_law_1.2,0.3076544046401978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,2,power_law_1.01,1.8871936798095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,2,power_law_1.2,0.4277440071105957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,2,power_law_1.01,1.318841552734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,2,1,power_law_1.2,2.9537984848022463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,2,power_law_1.01,2.7564672470092773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,2,1,power_law_1.2,15.882771301269532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,2,power_law_1.2,0.6561600208282471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,2,power_law_1.01,2.0895360946655273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,2,power_law_1.01,1.2858112335205079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,2,power_law_1.2,0.11589759588241577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,2,power_law_1.2,0.9495552062988282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,2,power_law_1.2,0.19972480535507203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,2,power_law_1.01,1.3544896125793457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,2,power_law_1.01,2.3180927276611327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,2,power_law_1.2,0.2977535963058472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,2,power_law_1.2,1.224409580230713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,2,power_law_1.01,1.4596735954284668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,2,power_law_1.2,0.43916797637939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,2,power_law_1.01,2.7343423843383787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,2,power_law_1.2,1.5545536041259767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,2,power_law_1.2,0.6294911861419678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,2,power_law_1.01,1.3432255744934083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,2,power_law_1.01,3.0879232406616213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,2,power_law_1.01,0.056524801254272464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,2,power_law_1.01,1.419545555114746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,2,power_law_1.01,0.08399999737739564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,2,power_law_1.01,3.2057281494140626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,2,power_law_1.01,0.1384384036064148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,2,power_law_1.01,1.5449407577514649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,2,power_law_1.2,0.10855040550231934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,2,power_law_1.01,3.166201591491699
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,2,power_law_1.01,0.17241599559783935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,2,power_law_1.2,0.1826367974281311
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,2,power_law_1.01,0.27842559814453127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,2,power_law_1.01,1.4920831680297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,2,power_law_1.2,0.30189440250396726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,2,power_law_1.2,0.7914048194885254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,2,power_law_1.01,0.3481663942337036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,2,power_law_1.2,1.863654327392578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,2,power_law_1.2,0.4269120216369629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,2,power_law_1.01,3.6461055755615233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,2,power_law_1.01,0.4031040191650391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,2,power_law_1.2,0.971615982055664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,2,power_law_1.2,0.6131392002105713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,2,power_law_1.01,0.5181503772735596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,2,power_law_1.2,1.9983104705810546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,2,power_law_1.2,1.2466112136840821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,2,power_law_1.2,0.8083135604858398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,2,power_law_1.01,0.5014783859252929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,2,1,power_law_1.2,10.565171051025391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,2,power_law_1.01,4.1308032989501955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,2,power_law_1.01,4.220755386352539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,2,power_law_1.2,1.9649471282958983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,2,power_law_1.01,0.5541952133178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,2,power_law_1.2,1.3544384002685548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,2,power_law_1.01,1.6150079727172852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,2,power_law_1.01,0.5444608211517334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,2,power_law_1.2,1.367801570892334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,2,power_law_1.2,2.051820755004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,2,power_law_1.01,0.5746560096740723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,2,power_law_1.01,1.6564735412597655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,2,power_law_1.01,4.995942306518555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,2,power_law_1.01,0.6009535789489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,2,power_law_1.01,5.137145614624023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,2,power_law_1.2,2.028268814086914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,2,power_law_1.01,1.7805440902709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,2,power_law_1.01,0.628223991394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,2,power_law_1.01,0.6600575923919678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,2,power_law_1.01,1.9513536453247071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,2,power_law_1.2,2.2746944427490234
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,2,power_law_1.01,0.7167168140411377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,2,power_law_1.01,0.6781184196472168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,2,power_law_1.01,6.089100646972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,2,power_law_1.01,2.246847915649414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,2,power_law_1.2,2.569286346435547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,2,power_law_1.01,0.7750336170196533
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,2,power_law_1.01,0.7629312038421631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,2,power_law_1.01,2.558700752258301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,2,power_law_1.2,2.6569536209106444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,2,power_law_1.01,0.8957887649536133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,2,power_law_1.2,1.3974847793579102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,2,power_law_1.01,6.682790374755859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,2,power_law_1.2,0.9536383628845215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,2,power_law_1.01,0.8803008079528809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,2,power_law_1.01,3.240185546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,2,power_law_1.2,2.6486207962036135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,2,power_law_1.2,1.5133248329162599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,2,power_law_1.01,8.40387191772461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,2,power_law_1.01,1.049830436706543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,2,power_law_1.2,1.2532928466796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,2,power_law_1.2,1.5145279884338378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,2,power_law_1.2,2.8044736862182615
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,2,power_law_1.2,0.05621119737625122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,2,power_law_1.2,1.1911487579345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,2,power_law_1.01,3.964992141723633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,2,power_law_1.2,0.08396160006523132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,2,power_law_1.2,1.6266496658325196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,2,power_law_1.2,0.12158080339431762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,2,power_law_1.2,1.2810751914978027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,2,power_law_1.2,3.029804801940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,2,power_law_1.2,0.17331839799880983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,2,power_law_1.01,8.584652709960938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,2,power_law_1.2,1.3150208473205567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,2,power_law_1.2,0.2585088014602661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,2,power_law_1.01,5.2031806945800785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,2,power_law_1.2,1.3916095733642577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,2,power_law_1.01,10.030470275878907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,2,power_law_1.2,1.4052672386169434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,2,power_law_1.2,1.723731231689453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,2,power_law_1.2,1.5492416381835938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,2,power_law_1.01,1.1575936317443847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,2,power_law_1.2,3.24139518737793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,2,power_law_1.01,6.705900573730469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,2,power_law_1.2,1.5690239906311034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,2,power_law_1.01,1.4313983917236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,2,power_law_1.01,12.044448089599609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,2,power_law_1.2,1.4829824447631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,2,power_law_1.2,3.737171173095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,2,power_law_1.01,1.6917184829711913
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,2,power_law_1.2,0.34124159812927246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,2,power_law_1.2,1.6710592269897462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,2,power_law_1.2,0.34612479209899905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,2,power_law_1.2,1.8302656173706056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,2,power_law_1.01,2.2226816177368165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,2,power_law_1.2,0.4811903953552246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,2,power_law_1.2,4.218406295776367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,2,power_law_1.01,9.288524627685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,2,power_law_1.2,2.0077888488769533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,2,power_law_1.2,1.847635269165039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,2,power_law_1.01,4.163308715820312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,2,power_law_1.2,2.304230308532715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,2,power_law_1.01,14.337971496582032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,2,power_law_1.2,5.315270233154297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,2,1,balanced,0.09963732957839966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,2,1,balanced,0.14338133732477823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,2,1,balanced,0.29363733530044556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,2,power_law_1.2,0.522540807723999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,2,1,balanced,0.5194880167643229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,2,power_law_1.2,2.6714431762695314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,2,1,balanced,0.9589440027872721
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,2,power_law_1.2,0.49622402191162107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,2,power_law_1.01,15.857676696777343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,2,1,balanced,1.8631092707316081
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,2,power_law_1.2,0.5501247882843018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,2,1,balanced,0.08533333738644917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,2,power_law_1.2,1.9315967559814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,2,1,balanced,0.1092639962832133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,2,1,balanced,0.1941866676012675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,2,power_law_1.2,3.3261505126953126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,2,power_law_1.2,0.5860928058624267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,2,power_law_1.2,6.3322303771972654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,2,1,balanced,2.7394399642944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,2,power_law_1.01,11.9268798828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,2,power_law_1.2,0.5630911827087403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,2,power_law_1.2,2.2419200897216798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,2,1,balanced,2.749349276224772
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,2,power_law_1.2,0.6409152030944825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,2,power_law_1.2,4.048031997680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,2,power_law_1.2,0.7145343780517578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,2,1,balanced,2.780917485555013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,2,power_law_1.2,2.4291711807250977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,2,power_law_1.2,0.7437695980072021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,2,1,balanced,0.3304640054702759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,2,1,balanced,2.79093329111735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,2,1,balanced,0.6346240043640137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,2,power_law_1.2,0.7084799766540527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,2,power_law_1.2,2.958982467651367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,2,1,balanced,1.2124319871266682
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,2,power_law_1.2,0.803660774230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,2,1,balanced,2.7970558802286782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,2,1,balanced,1.789903958638509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,2,power_law_1.2,5.316640090942383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,2,power_law_1.01,18.150405883789062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,2,power_law_1.2,0.7864384174346923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,2,1,balanced,2.8064746856689453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,2,1,balanced,1.7937599817911785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,2,power_law_1.2,3.332672119140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,2,power_law_1.2,0.9269568443298339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,2,1,balanced,1.7961173057556152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,2,1,balanced,2.837183952331543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,2,power_law_1.2,0.9442624092102051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,2,power_law_1.2,8.385139465332031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,2,power_law_1.2,1.1112704277038574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,2,1,balanced,2.8399572372436523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,2,power_law_1.2,4.197107315063477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,2,power_law_1.2,6.766297912597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,2,1,balanced,0.07832533121109009
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,2,power_law_1.2,1.3365632057189942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,2,1,balanced,0.10587732990582784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,2,1,balanced,0.18136000633239746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,2,1,balanced,0.3322666684786479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,2,1,balanced,0.6130933364232382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,2,power_law_1.2,1.4897024154663085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,2,1,balanced,1.7984639803568523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,2,1,balanced,1.1775733629862468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,2,power_law_1.2,5.145817565917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,2,1,balanced,1.805023988087972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,2,1,balanced,1.7426560719807942
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,2,power_law_1.2,1.8494911193847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,2,1,balanced,1.8104747136433919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,2,1,balanced,2.894517262776693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,2,power_law_1.2,10.847277069091797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,2,1,balanced,1.8251412709554036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,2,1,balanced,2.8367252349853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,2,1,balanced,1.8315199216206868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,2,1,balanced,1.8452746073404949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,2,power_law_1.2,7.024384307861328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,2,1,balanced,3.0419467290242515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,2,1,balanced,1.7484480539957683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,2,1,balanced,1.9218932787577312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,2,1,balanced,1.7525547345479329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,2,1,balanced,1.9876160621643066
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,2,power_law_1.2,2.3080703735351564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,2,1,balanced,1.760543982187907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,2,1,balanced,2.0522240002950034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,2,1,balanced,1.77127472559611
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,2,1,balanced,0.03764266769091288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,2,1,balanced,0.05826666454474131
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,2,1,balanced,0.08596799770991008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,2,1,balanced,2.164186636606852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,2,1,balanced,0.14174399773279825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,2,1,balanced,1.7818986574808757
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,2,1,balanced,0.24882133801778158
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,2,1,balanced,0.4641439914703369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,2,1,balanced,1.7947786649068196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,2,1,balanced,3.144282658894857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,2,power_law_1.2,9.247666931152343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,2,power_law_1.2,9.283302307128906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,2,1,balanced,1.8122186660766602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,2,power_law_1.2,14.989082336425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,2,1,balanced,3.181002616882324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,2,1,balanced,1.832032044728597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,2,1,balanced,1.8514240582784016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,2,1,balanced,3.3579839070638022
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,2,1,balanced,0.6850506464640299
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,2,power_law_1.2,4.4081470489501955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,2,1,balanced,1.8869226773579915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,2,1,balanced,0.6851360003153483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,2,1,balanced,0.6904853185017904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,2,1,balanced,1.9538559913635254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,2,1,balanced,3.5509332021077475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,2,1,balanced,0.6939360300699869
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,2,1,balanced,0.6991253693898519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,2,1,balanced,2.0177812576293945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,2,1,balanced,0.7057546774546305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,2,1,balanced,0.713269313176473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,2,1,balanced,3.905680020650228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,2,1,balanced,2.1254560152689614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,2,1,balanced,2.3401333491007485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,2,power_law_1.2,12.229126739501954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,2,power_law_1.2,12.049356842041016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,2,1,balanced,2.2438987096150718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,2,1,balanced,2.5474613507588706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,2,1,power_law_1.01,0.0971455991268158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,2,1,power_law_1.01,0.1405951976776123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,2,1,balanced,2.593183994293213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,2,1,power_law_1.01,0.08291199803352356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,2,1,balanced,2.9449278513590493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,2,1,power_law_1.01,0.29501440525054934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,2,1,power_law_1.01,0.1074944019317627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,2,1,balanced,0.7265600363413492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,2,1,power_law_1.01,0.5245823860168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,2,1,balanced,0.7405280272165934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,2,1,power_law_1.01,0.19328000545501708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,2,1,balanced,3.74069881439209
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,2,1,balanced,0.7540213267008463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,2,1,balanced,0.8144640127817789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,2,power_law_1.2,19.918975830078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,2,1,balanced,0.8649066289265951
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,2,1,balanced,0.8514986832936605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,2,1,balanced,4.274858792622884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,2,1,balanced,0.9241920312245687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,2,1,balanced,0.9856586456298828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,2,1,balanced,1.09170667330424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,2,1,balanced,4.369093259175618
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,2,1,balanced,1.4103199640909831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,2,1,balanced,5.818560282389323
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,2,1,balanced,1.3928052584330242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,2,1,balanced,5.642730712890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,2,power_law_1.2,16.79078369140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,2,1,balanced,2.983759880065918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,2,1,balanced,1.9496372540791829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,2,1,power_law_1.01,0.8782527923583985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,2,1,balanced,6.114896138509114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,2,1,balanced,7.923146565755208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,2,1,power_law_1.01,0.07617920041084289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,2,1,power_law_1.01,1.2370495796203613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,2,1,balanced,3.179957389831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,2,1,power_law_1.01,0.10397440195083618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,2,1,balanced,5.220591862996419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,2,1,power_law_1.01,0.17993600368499757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,2,1,power_law_1.01,1.430624008178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,2,1,power_law_1.2,0.09935359954833985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,2,1,power_law_1.01,0.3302079916000366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,2,1,power_law_1.01,0.3261568069458008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,2,1,power_law_1.01,1.8556608200073241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,2,1,balanced,8.940223693847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,2,1,power_law_1.01,0.5792704105377198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,2,1,power_law_1.01,0.8104767799377441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,2,1,balanced,7.051829020182292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,2,1,balanced,11.040453592936197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,2,1,power_law_1.01,1.9428224563598633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,2,1,power_law_1.01,0.957151985168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,2,1,power_law_1.2,0.14237439632415771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,2,1,power_law_1.01,2.0005823135375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,2,1,power_law_1.2,0.2851775884628296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,2,1,power_law_1.01,1.2170432090759278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,2,1,power_law_1.01,0.5877952098846435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,2,1,power_law_1.2,0.5096640110015869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,2,1,power_law_1.01,0.8022656440734863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,2,1,power_law_1.01,1.2729855537414552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,2,1,power_law_1.01,2.101862335205078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,2,1,power_law_1.2,0.7659520149230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,2,1,balanced,11.66387685139974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,2,1,balanced,9.79481569925944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,2,1,power_law_1.01,0.9698495864868164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,2,1,power_law_1.01,1.3024191856384277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,2,1,power_law_1.2,1.083948802947998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,2,1,power_law_1.01,2.1196224212646486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,2,1,power_law_1.2,0.08392959833145142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,2,1,power_law_1.01,1.2100607872009277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,2,1,power_law_1.01,1.3714624404907227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,2,1,power_law_1.2,0.1081536054611206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,2,1,power_law_1.2,1.3144191741943358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,2,1,power_law_1.2,0.1923583984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,2,1,power_law_1.01,2.2299711227416994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,2,1,power_law_1.01,1.424185562133789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,2,1,power_law_1.2,0.3272128105163574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,2,1,power_law_1.01,0.035795199871063235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,2,1,power_law_1.2,1.8219392776489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,2,1,power_law_1.2,0.5272448062896729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,2,1,power_law_1.01,1.4978303909301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,2,1,power_law_1.01,0.05724160075187683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,2,1,power_law_1.01,2.3976831436157227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,2,1,power_law_1.2,0.7182144165039063
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,2,1,power_law_1.01,0.08423039913177491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,2,1,power_law_1.01,1.6473600387573242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,2,1,power_law_1.2,1.883135986328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,2,1,balanced,13.580677032470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,2,1,power_law_1.2,0.838924789428711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,2,1,power_law_1.01,1.261030387878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,2,1,power_law_1.01,2.4862272262573244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,2,1,power_law_1.01,1.7994495391845704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,2,1,power_law_1.2,1.9613887786865234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,2,1,power_law_1.2,1.1846783638000489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,2,1,power_law_1.01,1.3371711730957032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,2,1,power_law_1.01,2.6840063095092774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,2,1,power_law_1.01,1.924006462097168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,2,1,power_law_1.2,1.2583488464355468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,2,1,power_law_1.2,2.0208831787109376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,2,1,power_law_1.01,1.366105556488037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,2,1,power_law_1.01,0.13933440446853637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,2,1,power_law_1.01,1.7731584548950194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,2,1,power_law_1.01,2.661222457885742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,2,1,power_law_1.01,1.4034624099731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,2,1,power_law_1.01,0.22833919525146484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,2,1,power_law_1.2,2.1289087295532227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,2,1,power_law_1.01,0.3256767988204956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,2,1,power_law_1.01,1.4498047828674316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,2,1,power_law_1.01,1.9838848114013672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,2,1,balanced,19.8538080851237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,2,1,power_law_1.01,0.39582080841064454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,2,1,power_law_1.01,2.8546688079833986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,2,1,power_law_1.2,2.225881576538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,2,1,power_law_1.01,1.4804160118103027
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,2,1,power_law_1.01,0.4841343879699707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,2,1,power_law_1.01,2.1141952514648437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,2,1,power_law_1.2,0.07692800164222717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,2,1,power_law_1.2,1.2716992378234864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,2,1,power_law_1.01,0.5071807861328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,2,1,power_law_1.2,2.3880640029907227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,2,1,power_law_1.01,2.977529525756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,2,1,power_law_1.01,0.5297664165496826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,2,1,power_law_1.2,1.3522239685058595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,2,1,power_law_1.01,2.5102720260620117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,2,1,power_law_1.01,0.5462719917297363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,2,1,power_law_1.2,2.563283157348633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,2,1,power_law_1.2,1.4375359535217285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,2,1,power_law_1.01,0.5593344211578369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,2,1,power_law_1.01,3.4061630249023436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,2,1,power_law_1.01,2.872960090637207
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,2,1,power_law_1.01,0.5835775852203369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,2,1,power_law_1.2,1.5247039794921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,2,1,power_law_1.2,0.10517120361328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,2,1,power_law_1.2,2.8115776062011717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,2,1,power_law_1.2,0.17997440099716186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,2,1,power_law_1.2,1.6670143127441406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,2,1,power_law_1.01,3.8657150268554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,2,1,power_law_1.2,0.3301568031311035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,2,1,power_law_1.01,3.6252159118652343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,2,1,power_law_1.2,2.670297622680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,2,1,power_law_1.2,0.5303167819976806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,2,1,power_law_1.2,1.836511993408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,2,1,power_law_1.2,0.7194687843322753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,2,1,power_law_1.2,2.8777984619140624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,2,1,power_law_1.2,1.9622976303100585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,2,1,power_law_1.01,4.40252799987793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,2,1,power_law_1.2,0.9195839881896972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,2,1,power_law_1.2,1.8066495895385741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,2,1,power_law_1.2,1.1836928367614745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,2,1,power_law_1.2,3.1201663970947267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,2,1,power_law_1.01,1.5225536346435546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,2,1,power_law_1.2,2.011769676208496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,2,1,power_law_1.2,1.2513728141784668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,2,1,power_law_1.01,0.606771183013916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,2,1,power_law_1.01,4.775481414794922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,2,1,power_law_1.01,5.915161514282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,2,1,power_law_1.01,1.5727423667907714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,2,1,power_law_1.2,3.4613567352294923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,2,1,power_law_1.01,0.6387135982513428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,2,1,power_law_1.2,2.1752191543579102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,2,1,power_law_1.2,0.03570559918880463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,2,1,power_law_1.01,0.6726143836975098
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,2,1,power_law_1.2,0.057132798433303836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,2,1,power_law_1.01,1.6089599609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,2,1,power_law_1.01,0.7411647796630859
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,2,1,power_law_1.2,0.08464639782905578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,2,1,power_law_1.2,2.6105472564697267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,2,1,power_law_1.2,3.9318336486816405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,2,1,power_law_1.01,5.560556793212891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,2,1,power_law_1.01,1.7850048065185546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,2,1,power_law_1.2,1.2588864326477052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,2,1,power_law_1.01,7.4596412658691404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,2,1,power_law_1.2,2.978451156616211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,2,1,power_law_1.01,1.919206428527832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,2,1,power_law_1.2,1.3644351959228516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,2,1,power_law_1.2,5.084044647216797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,2,1,power_law_1.01,2.166624069213867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,2,1,power_law_1.2,0.13861119747161865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,2,1,power_law_1.01,0.9285120010375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,2,1,power_law_1.2,3.776454544067383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,2,1,power_law_1.2,0.20909440517425537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,2,1,power_law_1.01,7.154259490966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,2,1,power_law_1.01,0.8242176055908204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,2,1,power_law_1.01,2.4503103256225587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,2,1,power_law_1.2,0.2957312107086182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,2,1,power_law_1.01,1.0225536346435546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,2,1,power_law_1.2,0.36394240856170657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,2,1,power_law_1.2,5.835084915161133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,2,1,power_law_1.2,4.520915222167969
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,2,1,power_law_1.01,0.9779583930969238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,2,1,power_law_1.2,1.366483211517334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,2,1,power_law_1.2,0.47411198616027833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,2,1,power_law_1.01,10.560364532470704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,2,1,power_law_1.01,3.0695552825927734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,2,1,power_law_1.01,1.2306495666503907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,2,1,power_law_1.2,0.49190402030944824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,2,1,power_law_1.2,1.4402432441711426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,2,1,power_law_1.2,0.5004608154296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,2,1,power_law_1.01,1.2270400047302246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,2,1,power_law_1.01,9.122361755371093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,2,1,power_law_1.01,3.6715839385986326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,2,1,power_law_1.2,0.5365568161010742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,2,1,power_law_1.2,6.104889678955078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,2,1,power_law_1.01,1.5652928352355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,2,1,power_law_1.2,0.5515711784362793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,2,1,power_law_1.2,0.5767680168151855
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,2,1,power_law_1.01,1.6807039260864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,2,1,power_law_1.2,0.5988224029541016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,2,1,power_law_1.01,4.860115051269531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,2,1,power_law_1.2,1.4324864387512206
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,2,1,power_law_1.2,0.6533567905426025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,2,1,power_law_1.01,2.2345344543457033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,2,1,power_law_1.2,7.675385284423828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,2,1,power_law_1.2,1.5423935890197753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,2,1,power_law_1.2,7.636332702636719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,2,1,power_law_1.2,1.5685183525085449
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,2,1,power_law_1.01,3.4707584381103516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,2,1,power_law_1.01,13.130233764648438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,2,balanced,0.060496002435684204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,2,1,power_law_1.2,1.6652032852172851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,2,balanced,0.07448533177375793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,2,balanced,0.10643200079600017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,2,balanced,0.1883253256479899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,2,balanced,0.3381919860839844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,2,1,power_law_1.2,0.6871103763580322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,2,balanced,0.600927988688151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,2,1,power_law_1.01,13.643373107910156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,2,balanced,0.6032640139261881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,2,1,power_law_1.2,1.8128320693969726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,2,1,power_law_1.2,0.7715839862823486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,2,balanced,0.043866669138272606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,2,balanced,0.6067626476287842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,2,balanced,0.06609599788983662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,2,balanced,0.09778133034706116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,2,balanced,0.6091253360112509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,2,balanced,0.14072533448537192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,2,1,power_law_1.2,9.235295867919922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,2,balanced,0.611573338508606
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,2,1,power_law_1.2,0.9741951942443847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,2,balanced,0.6180213292439779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,2,1,power_law_1.01,6.041939163208008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,2,balanced,0.7106133302052816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,2,1,power_law_1.2,0.8682559967041016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,2,balanced,0.7161280314127604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,2,1,power_law_1.2,10.74247055053711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,2,balanced,0.7291413148244222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,2,balanced,0.7397066752115885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,2,balanced,0.7511733373006185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,2,balanced,0.7731520334879557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,2,balanced,0.23162132501602173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,2,balanced,0.8240533669789633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,2,balanced,0.41652798652648926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,2,1,power_law_1.2,1.960767936706543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,2,balanced,0.8727893034617106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,2,balanced,0.96234663327535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,2,balanced,1.0600266456604004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,2,1,power_law_1.01,8.458751678466797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,2,1,power_law_1.2,1.0837696075439454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,2,balanced,1.454367955525716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,2,1,power_law_1.01,16.607781982421876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,2,1,power_law_1.2,1.039129638671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,2,balanced,1.680239995320638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,2,1,power_law_1.2,12.866732788085937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,2,balanced,0.4175146818161011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,2,balanced,0.4189066489537557
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,2,1,power_law_1.2,1.321395206451416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,2,balanced,0.4220159848531087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,2,balanced,0.42371201515197754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,2,balanced,0.054976001381874084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,2,balanced,0.4280159870783488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,2,1,power_law_1.2,2.258815956115723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,2,1,power_law_1.2,13.863679504394531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,2,1,power_law_1.2,2.5804672241210938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,2,balanced,2.349082628885905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,2,1,power_law_1.01,10.810233306884765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,2,balanced,0.06258133550484975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,2,power_law_1.01,0.06756479740142822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,2,balanced,0.4657440185546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,2,balanced,3.014399846394857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,2,balanced,0.47253334522247314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,2,power_law_1.01,0.09968000054359435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,2,1,power_law_1.2,3.1676544189453124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,2,balanced,0.4875253438949585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,2,balanced,0.4990239938100179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,2,balanced,0.5123733282089233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,2,balanced,0.5386720101038615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,2,balanced,0.5993066628774008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,2,balanced,4.8141279220581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,2,balanced,0.6490453481674194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,2,1,power_law_1.2,3.7529022216796877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,2,balanced,0.029285334050655365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,2,balanced,0.03362133353948593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,2,balanced,0.09674666325251262
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,2,balanced,0.049584001302719116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,2,1,power_law_1.2,1.2781375885009765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,2,balanced,0.07003733515739441
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,2,balanced,0.10743467013041179
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,2,balanced,0.17388800779978433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,2,balanced,0.17673067251841226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,2,balanced,0.17875200510025024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,2,balanced,0.18040533860524496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,2,balanced,0.1827359994252523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,2,1,power_law_1.2,1.6243263244628907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,2,balanced,0.1868106722831726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,2,balanced,0.18997333447138467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,2,balanced,6.76962153116862
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,2,balanced,0.1936639944712321
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,2,balanced,0.1995733380317688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,2,balanced,0.19475199778874716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,2,1,power_law_1.2,4.951283264160156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,2,balanced,0.19894399245580038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,2,balanced,0.7607413132985433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,2,balanced,0.20839466651280722
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,2,1,power_law_1.2,1.8300352096557617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,2,balanced,0.8770666917165121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,2,1,power_law_1.2,16.68035888671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,2,balanced,1.294922669728597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,2,power_law_1.01,0.17274240255355836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,2,balanced,1.5389973322550456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,2,power_law_1.01,0.23315200805664063
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,2,1,power_law_1.2,2.325766372680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,2,power_law_1.01,0.30485761165618896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,2,power_law_1.01,0.05975040197372437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,2,power_law_1.01,0.06183680295944214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,2,power_law_1.01,0.352512001991272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,2,power_law_1.01,0.09349120259284974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,2,power_law_1.01,0.09499520063400269
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,2,balanced,0.20972265799840292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,2,balanced,0.219866673151652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,2,power_law_1.01,0.11159679889678956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,2,power_law_1.01,0.11600639820098876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,2,power_law_1.01,0.49781122207641604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,2,power_law_1.01,0.16380159854888915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,2,1,power_law_1.2,3.679993438720703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,2,power_law_1.01,0.4675583839416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,2,power_law_1.01,0.21656959056854247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,2,power_law_1.01,0.5203584194183349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,2,power_law_1.01,0.2838207960128784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,2,balanced,2.208639939626058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,2,power_law_1.01,0.3486848115921021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,2,1,power_law_1.2,6.134067153930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,2,power_law_1.01,0.3701567888259888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,2,balanced,0.137445330619812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,2,balanced,0.23819732666015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,2,balanced,2.8814026514689126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,2,power_law_1.01,0.36935040950775144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,2,balanced,0.43078935146331787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,2,power_law_1.01,0.15463680028915405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,2,balanced,0.43556265036265057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,2,balanced,0.43855468432108563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,2,power_law_1.01,0.20914559364318847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,2,balanced,0.4420320192972819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,2,balanced,0.44467735290527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,2,power_law_1.01,0.2446847915649414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,2,balanced,0.4527093172073364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,2,balanced,4.235616048177083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,2,balanced,0.45888535181681317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,2,power_law_1.01,0.5265408039093018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,2,power_law_1.01,0.3181823968887329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,2,balanced,0.4651626745859782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,2,balanced,0.47764265537261963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,2,power_law_1.01,0.3257791996002197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,2,balanced,0.492304007212321
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,2,balanced,0.23212265968322754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,2,balanced,0.5050613482793173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,2,power_law_1.01,0.32584319114685056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,2,balanced,0.24462399880091348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,2,balanced,0.528661330540975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,2,power_law_1.01,0.029951998591423036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,2,balanced,0.33643198013305664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,2,power_law_1.01,0.3534656047821045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,2,balanced,0.57860799630483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,2,power_law_1.01,0.36690559387207033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,2,balanced,5.586655934651692
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,2,power_law_1.01,0.04604159891605377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,2,balanced,0.6220053434371948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,2,power_law_1.01,0.3709439992904663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,2,power_law_1.01,0.3987071990966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,2,power_law_1.2,0.06801279783248901
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,2,power_law_1.01,0.41341438293457033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,2,power_law_1.01,0.4269887924194336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,2,power_law_1.2,0.09995520114898682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,2,power_law_1.01,0.41637120246887205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,2,power_law_1.2,0.15574400424957274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,2,power_law_1.01,0.5275008201599121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,2,power_law_1.01,0.4416192054748535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,2,power_law_1.2,0.21877760887145997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,2,power_law_1.01,0.60316162109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,2,power_law_1.01,0.45354881286621096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,2,power_law_1.2,0.2747391939163208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,2,balanced,0.29783467451731366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,2,1,power_law_1.2,8.501344299316406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,2,power_law_1.01,0.47260160446166993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,2,power_law_1.01,0.6125951766967773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,2,power_law_1.2,0.3320832014083862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,2,power_law_1.01,0.5206208229064941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,2,balanced,0.7114933331807455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,2,power_law_1.01,0.6455423831939697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,2,balanced,0.8257760206858317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,2,power_law_1.01,0.6094016075134278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,2,power_law_1.01,0.6648128032684326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,2,power_law_1.01,0.41069440841674804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,2,balanced,1.1961599985758464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,2,power_law_1.01,0.7040256023406982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,2,power_law_1.01,0.4450240135192871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,2,power_law_1.01,0.6786367893218994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,2,balanced,1.402773380279541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,2,power_law_1.01,0.867033576965332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,2,power_law_1.01,0.4525951862335205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,2,power_law_1.01,0.7375360012054444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,2,balanced,1.9897546768188477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,2,balanced,0.5882186492284139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,2,power_law_1.01,1.0379903793334961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,2,balanced,0.5034720102945963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,2,power_law_1.01,0.8052800178527832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,2,power_law_1.01,0.05428479909896851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,2,power_law_1.2,0.45731201171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,2,balanced,0.7171626885732015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,2,power_law_1.01,0.07391999959945679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,2,power_law_1.01,1.369273567199707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,2,power_law_1.2,0.48295040130615235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,2,balanced,0.9299253622690836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,2,power_law_1.01,0.09719039797782898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,2,power_law_1.2,0.49699840545654295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,2,power_law_1.01,0.11726720333099365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,2,power_law_1.01,1.717625617980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,2,1,power_law_1.2,10.87570571899414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,2,power_law_1.01,0.14158719778060913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,2,power_law_1.2,0.5113344192504883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,2,power_law_1.01,0.4621888160705566
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,2,power_law_1.01,0.14877439737319947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,2,power_law_1.2,0.5536255836486816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,2,power_law_1.01,0.5066239833831787
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,2,power_law_1.01,0.15303679704666137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,2,power_law_1.01,2.500217628479004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,2,balanced,2.604464054107666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,2,power_law_1.01,0.5884160041809082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,2,power_law_1.2,0.06020479798316956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,2,power_law_1.01,0.8581695556640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,2,power_law_1.2,0.09245439767837524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,2,power_law_1.01,0.6583231925964356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,2,power_law_1.2,0.06103039979934692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,2,power_law_1.01,1.0262911796569825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,2,power_law_1.2,0.11127040386199952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,2,power_law_1.2,0.09167360067367554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,2,power_law_1.01,0.801030445098877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,2,balanced,3.790442784627279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,2,power_law_1.2,0.1687999963760376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,2,power_law_1.2,0.10798079967498779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,2,power_law_1.01,1.2163840293884278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,2,power_law_1.01,0.9590463638305664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,2,power_law_1.2,0.1885632038116455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,2,power_law_1.2,0.2757055997848511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,2,power_law_1.01,1.2439743995666503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,2,power_law_1.2,0.36278400421142576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,2,power_law_1.2,0.35735039710998534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,2,power_law_1.01,3.189900779724121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,2,power_law_1.2,0.37586560249328616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,2,power_law_1.2,0.3747904062271118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,2,power_law_1.2,0.3933759927749634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,2,power_law_1.01,1.5440959930419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,2,power_law_1.2,0.6040256023406982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,2,power_law_1.2,0.41271038055419923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,2,balanced,4.968165397644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,2,power_law_1.2,0.6085055828094482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,2,power_law_1.2,0.4139840126037598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,2,power_law_1.01,4.631193542480469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,2,power_law_1.2,0.650707197189331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,2,power_law_1.2,0.43695359230041503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,2,power_law_1.2,0.7083392143249512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,2,power_law_1.2,0.4670720100402832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,2,power_law_1.2,0.14731520414352417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,2,power_law_1.2,0.7038720130920411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,2,power_law_1.2,0.47277441024780276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,2,power_law_1.2,0.19116159677505493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,2,power_law_1.2,0.7437183856964111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,2,power_law_1.2,0.24186880588531495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,2,power_law_1.2,0.827564811706543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,2,power_law_1.2,0.31100800037384035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,2,power_law_1.01,1.8652416229248048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,2,power_law_1.01,5.9265281677246096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,2,power_law_1.2,0.3316159963607788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,2,power_law_1.2,0.9090111732482911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,2,power_law_1.2,0.34023680686950686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,2,power_law_1.2,1.0889151573181153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,2,power_law_1.2,0.3517312049865723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,2,power_law_1.01,2.5926336288452148
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,2,power_law_1.2,0.029190400242805482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,2,power_law_1.01,0.15159679651260377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,2,power_law_1.2,1.231827163696289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,2,power_law_1.2,0.04455040097236633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,2,power_law_1.01,0.1635648012161255
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,2,power_law_1.2,0.053497600555419925
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,2,power_law_1.01,0.16706559658050538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,2,power_law_1.01,1.5527680397033692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,2,power_law_1.2,1.5894975662231445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,2,power_law_1.01,3.287667083740234
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,2,power_law_1.01,0.17221120595932007
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,2,power_law_1.2,0.07212160229682922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,2,power_law_1.01,0.18764159679412842
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,2,power_law_1.2,0.09466239809989929
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,2,power_law_1.01,0.1788800001144409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,2,power_law_1.2,1.9636032104492187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,2,power_law_1.2,0.1074112057685852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,2,power_law_1.01,0.18320640325546264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,2,power_law_1.2,0.3701567888259888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,2,power_law_1.01,0.19930880069732665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,2,power_law_1.2,0.42332801818847654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,2,power_law_1.01,0.20346879959106445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,2,power_law_1.01,4.767507171630859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,2,power_law_1.2,2.654342460632324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,2,power_law_1.01,0.22828800678253175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,2,1,balanced,0.04762666424115499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,2,1,balanced,0.0621973325808843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,2,1,balanced,0.09759466846783955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,2,power_law_1.01,0.23724799156188964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,2,power_law_1.01,0.2717695951461792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,2,power_law_1.01,2.1900608062744142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,2,power_law_1.01,0.3442176103591919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,2,power_law_1.2,3.448863983154297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,2,power_law_1.2,0.5316224098205566
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,2,power_law_1.01,0.3990272045135498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,2,power_law_1.2,0.5991487979888916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,2,power_law_1.01,2.7886911392211915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,2,power_law_1.01,6.1447296142578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,2,power_law_1.2,0.6618688106536865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,2,1,balanced,0.17504000663757324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,2,1,balanced,0.32840534051259357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,2,power_law_1.2,0.8315839767456055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,2,1,balanced,0.5985546509424845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,2,power_law_1.2,4.917055892944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,2,1,balanced,0.6006293296813965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,2,power_law_1.2,0.9962688446044922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,2,power_law_1.2,0.14416639804840087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,2,power_law_1.01,3.984505462646484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,2,power_law_1.2,0.14922239780426025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,2,power_law_1.2,1.3198911666870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,2,power_law_1.01,0.505075216293335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,2,power_law_1.2,0.14325759410858155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,2,1,balanced,0.04148799926042557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,2,1,balanced,0.055120001236597695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,2,power_law_1.01,0.6179456233978271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,2,power_law_1.2,1.5684736251831055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,2,power_law_1.01,0.8862015724182128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,2,power_law_1.2,6.449260711669922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,2,power_law_1.01,5.227449417114258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,2,1,balanced,0.60152534643809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,2,1,balanced,0.6040800015131632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,2,power_law_1.2,2.2198207855224608
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,2,power_law_1.01,1.434879970550537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,2,power_law_1.2,0.4288959980010986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,2,1,power_law_1.01,0.04615679979324341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,2,power_law_1.2,0.4438015937805176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,2,1,power_law_1.01,0.05997440218925476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,2,1,balanced,0.08834133545557658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,2,1,balanced,0.04214400053024292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,2,1,balanced,0.1341493328412374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,2,1,balanced,0.05884799857934316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,2,1,balanced,0.09062400460243225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,2,power_law_1.2,0.4670015811920166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,2,1,power_law_1.01,0.0942911982536316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,2,power_law_1.2,0.4953919887542725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,2,power_law_1.2,0.5411968231201172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,2,1,balanced,0.6077066659927368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,2,power_law_1.2,0.6332608222961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,2,1,balanced,0.6099573373794556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,2,1,balanced,0.6146506468454996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,2,power_law_1.2,0.7206719875335693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,2,1,balanced,0.6201973358790079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,2,1,balanced,0.6242133378982544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,2,1,balanced,0.23614400625228882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,2,power_law_1.2,2.884998321533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,2,power_law_1.2,0.8844799995422363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,2,1,balanced,0.6919999917348226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,2,1,power_law_1.01,0.17488640546798706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,2,power_law_1.2,1.0579327583312987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,2,1,power_law_1.01,0.25075840950012207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,2,1,power_law_1.01,0.3228480100631714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,2,power_law_1.2,1.4132927894592284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,2,power_law_1.2,3.9746047973632814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,2,power_law_1.2,0.15730559825897217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,2,power_law_1.2,0.16340479850769044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,2,1,balanced,0.13796266913414001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,2,1,balanced,0.42692800362904865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,2,1,balanced,0.24966933329900107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,2,1,balanced,0.6972906589508057
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,2,power_law_1.2,0.17003519535064698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,2,1,balanced,0.42973868052164715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,2,1,balanced,0.7112053235371908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,2,1,balanced,0.4309759934743245
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,2,power_law_1.2,0.178438401222229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,2,1,balanced,0.4306666851043701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,2,1,balanced,0.73963729540507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,2,1,balanced,0.43395201365152997
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,2,power_law_1.2,0.1925696015357971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,2,1,balanced,0.7675253550211588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,2,1,balanced,0.43515201409657794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,2,1,balanced,0.4362186590830485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,2,1,balanced,0.8257866700490316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,2,1,power_law_1.01,0.44510722160339355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,2,power_law_1.2,5.519993591308594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,2,1,balanced,0.02757333219051361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,2,1,balanced,0.8940587043762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,2,1,power_law_1.01,0.4679103851318359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,2,1,power_law_1.01,0.033107200264930726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,2,1,balanced,1.2158666451772053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,2,power_law_1.2,1.8596031188964843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,2,1,power_law_1.01,0.48000640869140626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,2,1,balanced,0.44722668329874676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,2,1,power_law_1.01,0.49573121070861814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,2,1,balanced,0.4493120114008586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,2,1,balanced,0.45210667451222736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,2,1,power_law_1.01,0.5132736206054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,2,1,balanced,0.4527253309885661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,2,power_law_1.2,2.5678592681884767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,2,1,balanced,0.456005334854126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,2,1,power_law_1.01,0.5447679996490479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,2,1,balanced,0.43860801060994464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,2,1,balanced,0.0315733328461647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,2,1,power_law_1.01,0.5565887928009033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,2,1,balanced,0.04597333570321401
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,2,1,balanced,0.06630933284759521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,2,1,balanced,0.1036906639734904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,2,1,balanced,0.17350933949152628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,2,1,power_law_1.01,0.6152575969696045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,2,1,balanced,0.17557867368062338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,2,power_law_1.2,3.2464576721191407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,2,1,balanced,1.3676800727844238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,2,1,power_law_1.01,0.6216512203216553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,2,1,power_law_1.01,0.6488512039184571
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,2,power_law_1.2,0.17487360239028932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,2,1,power_law_1.01,0.6697919845581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,2,power_law_1.2,0.188428795337677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,2,1,balanced,0.44545066356658936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,2,1,power_law_1.01,0.7175807952880859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,2,1,balanced,0.45052266120910645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,2,power_law_1.2,0.20804479122161865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,2,power_law_1.2,4.660172653198242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,2,1,balanced,0.4594133297602336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,2,power_law_1.2,0.21121280193328856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,2,1,balanced,0.4726826747258504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,2,1,balanced,0.1790026624997457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,2,1,balanced,0.5124373435974121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,2,power_law_1.2,0.23991041183471679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,2,1,balanced,0.5476160049438477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,2,power_law_1.2,0.2531647920608521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,2,1,balanced,0.6261226733525594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,2,1,balanced,1.8946720759073894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,2,1,balanced,0.7029706637064616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,2,power_law_1.2,0.27962241172790525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,2,1,balanced,0.46056000391642254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,2,1,balanced,1.058784008026123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,2,1,balanced,0.46511467297871906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,2,power_law_1.2,0.37366399765014646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,2,1,balanced,2.4269919395446777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,2,1,balanced,0.4659680128097534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,2,1,balanced,1.230394681294759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,2,1,balanced,0.4768799940745036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,2,power_law_1.2,0.413267183303833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,2,power_law_1.2,6.159494400024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,2,1,balanced,0.484005331993103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,2,1,power_law_1.01,0.055929601192474365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,2,1,power_law_1.01,0.7717631816864013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,2,1,balanced,0.4922666549682617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,2,1,balanced,1.764453411102295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,2,1,balanced,0.18073066075642905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,2,1,balanced,0.5056053400039673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,2,1,power_law_1.01,0.08638079762458802
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,2,1,balanced,0.1808906594912211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,2,1,balanced,0.5326879819234213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,2,1,balanced,0.1832746664683024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,2,1,balanced,0.1876159906387329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,2,1,power_law_1.01,0.1324735999107361
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,2,1,balanced,0.1902559995651245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,2,1,balanced,2.299066702524821
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,2,1,balanced,0.19545066356658936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,2,1,balanced,0.22233066956202188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,2,1,balanced,0.22569066286087036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,2,1,balanced,0.2394933303197225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,2,1,power_law_1.01,0.037939199805259706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,2,1,balanced,3.375450770060221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,2,1,power_law_1.01,0.057145601511001586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,2,1,power_law_1.01,0.08821759819984436
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,2,power_law_1.2,0.547603178024292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,2,1,balanced,3.586730639139811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,2,1,power_law_1.01,0.1366528034210205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,2,1,power_law_1.01,0.8832320213317871
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,2,power_law_1.2,0.6583040237426758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,2,1,balanced,0.5577013492584229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,2,1,power_law_1.01,0.20106239318847657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,2,1,balanced,4.452010790506999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,2,1,balanced,0.6110666592915853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,2,1,power_law_1.01,1.0097344398498536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,2,1,power_law_1.01,0.1844480037689209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,2,1,power_law_1.01,0.2619136095046997
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,2,power_law_1.2,0.8955455780029297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,2,1,power_law_1.01,0.23710079193115235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,2,1,balanced,0.23969600598017374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,2,1,power_law_1.01,0.34226560592651367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,2,1,balanced,4.5857547124226885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,2,1,balanced,0.2526879906654358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,2,1,power_law_1.01,0.31538560390472414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,2,1,power_law_1.01,0.35092480182647706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,2,1,power_law_1.2,0.046367999911308286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,2,power_law_1.2,1.6539199829101563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,2,1,power_law_1.01,0.3371263980865479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,2,1,power_law_1.2,0.06209920048713684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,2,1,power_law_1.2,0.03255040049552917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,2,1,power_law_1.01,0.3369663953781128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,2,1,power_law_1.2,0.09534080028533935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,2,1,power_law_1.2,0.053932797908782956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,2,1,power_law_1.2,0.16512000560760498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,2,1,power_law_1.01,0.347161602973938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,2,1,power_law_1.2,0.08559359908103943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,2,1,balanced,0.6871093114217123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,2,1,power_law_1.01,0.36620159149169923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,2,1,power_law_1.2,0.13057279586791992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,2,1,balanced,1.025877316792806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,2,1,power_law_1.2,0.1674880027770996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,2,1,power_law_1.01,0.4033792018890381
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,2,1,balanced,0.2773386637369792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,2,1,balanced,1.1510506470998128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,2,1,balanced,0.29791466395060223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,2,1,power_law_1.01,0.4142399787902832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,2,1,balanced,0.4034186601638794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,2,1,power_law_1.01,0.3682368040084839
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,2,1,balanced,0.38015464941660565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,2,1,power_law_1.01,0.47079038619995117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,2,1,balanced,0.8116426467895508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,2,1,power_law_1.01,0.3738303899765015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,2,1,power_law_1.01,0.41460480690002444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,2,1,balanced,0.5420373280843099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,2,1,power_law_1.01,0.39435520172119143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,2,1,power_law_1.01,0.4306367874145508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,2,1,balanced,0.7668480078379313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,2,1,power_law_1.2,0.23491840362548827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,2,1,power_law_1.01,0.40555520057678224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,2,1,power_law_1.01,0.46286721229553224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,2,1,power_law_1.2,0.3034303903579712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,2,1,power_law_1.01,1.2763456344604491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,2,1,power_law_1.01,0.40429439544677737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,2,1,power_law_1.01,0.5248064041137696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,2,1,power_law_1.2,0.2224128007888794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,2,1,power_law_1.01,0.42766079902648924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,2,1,power_law_1.01,0.5870975971221923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,2,1,power_law_1.01,0.02467840015888214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,2,1,power_law_1.01,1.529043197631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,2,1,balanced,1.643173376719157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,2,1,power_law_1.01,0.02940160036087036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,2,1,power_law_1.01,0.7143167972564697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,2,1,power_law_1.01,0.045542401075363156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,2,1,balanced,2.1229440371195474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,2,1,power_law_1.01,0.8441920280456543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,2,1,power_law_1.01,2.0734207153320314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,2,1,power_law_1.01,0.06504960060119629
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,2,1,balanced,0.9927626450856527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,2,1,power_law_1.01,0.08641279935836792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,2,1,balanced,3.091775894165039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,2,1,power_law_1.2,0.4359231948852539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,2,1,power_law_1.01,2.643507194519043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,2,1,power_law_1.2,0.31570560932159425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,2,1,power_law_1.2,0.46287999153137205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,2,1,power_law_1.2,0.32729599475860593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,2,1,power_law_1.2,0.48007678985595703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,2,1,balanced,4.058992067972819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,2,1,power_law_1.2,0.03772160112857818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,2,1,power_law_1.2,0.34104959964752196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,2,1,power_law_1.2,0.05720319747924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,2,1,power_law_1.2,0.3435391902923584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,2,1,power_law_1.2,0.0883903980255127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,2,1,power_law_1.01,1.1153663635253905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,2,1,power_law_1.01,3.9758655548095705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,2,1,power_law_1.2,0.38650240898132326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,2,1,power_law_1.2,0.13216639757156373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,2,1,power_law_1.01,0.11089279651641845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,2,1,power_law_1.2,0.4116032123565674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,2,1,power_law_1.2,0.1793984055519104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,2,1,power_law_1.01,1.3755200386047364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,2,1,power_law_1.01,0.43884801864624023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,2,1,power_law_1.2,0.4393599987030029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,2,1,power_law_1.2,0.24394240379333496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,2,1,power_law_1.01,0.4554751873016357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,2,1,power_law_1.2,0.487718391418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,2,1,power_law_1.2,0.3344575881958008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,2,1,power_law_1.01,0.4759103775024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,2,1,power_law_1.2,0.42960638999938966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,2,1,power_law_1.01,1.9179840087890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,2,1,power_law_1.2,0.5018496036529541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,2,1,power_law_1.2,0.35141758918762206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,2,1,power_law_1.01,0.5351295948028565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,2,1,power_law_1.2,0.4415103912353516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,2,1,power_law_1.2,0.5170559883117676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,2,1,power_law_1.2,0.36156160831451417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,2,1,power_law_1.2,0.4748544216156006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,2,1,power_law_1.2,0.5499904155731201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,2,1,power_law_1.2,0.3758784055709839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,2,1,power_law_1.01,2.4490047454833985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,2,1,power_law_1.2,0.54269437789917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,2,1,power_law_1.01,0.1392192006111145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,2,1,power_law_1.2,0.5616767883300782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,2,1,power_law_1.2,0.38908159732818604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,2,1,power_law_1.2,0.6101247787475585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,2,1,power_law_1.01,0.14771840572357178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,2,1,power_law_1.2,0.6332159996032715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,2,1,power_law_1.2,0.3993407964706421
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,2,1,power_law_1.01,0.14953600168228148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,2,1,power_law_1.2,0.7430399894714356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,2,1,power_law_1.2,0.4120192050933838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,2,1,power_law_1.2,0.6418303966522216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,2,1,power_law_1.01,0.15276800394058226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,2,1,power_law_1.01,3.528915023803711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,2,1,power_law_1.2,0.867091178894043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,2,1,power_law_1.2,0.6591551780700684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,2,1,power_law_1.01,5.12023696899414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,2,balanced,0.058245331048965454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,2,1,power_law_1.01,0.5876480102539062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,2,balanced,0.07393066585063934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,2,balanced,0.10502400000890096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,2,1,power_law_1.2,1.1412287712097169
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,2,1,power_law_1.2,0.027807998657226562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,2,balanced,0.18685867389043173
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,2,1,power_law_1.2,0.031167998909950256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,2,balanced,0.3378613392512004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,2,1,power_law_1.01,0.7124671936035156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,2,balanced,0.6049919923146566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,2,1,power_law_1.2,1.4020544052124024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,2,1,power_law_1.2,0.04704639911651611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,2,balanced,0.6073866685231527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,2,1,power_law_1.01,0.8270848274230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,2,1,power_law_1.2,0.06085119843482971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,2,balanced,0.6084640026092529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,2,1,power_law_1.01,4.608755111694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,2,1,power_law_1.2,0.0806335985660553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,2,1,power_law_1.2,0.424179220199585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,2,1,power_law_1.01,0.15975680351257324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,2,1,power_law_1.2,0.10442880392074586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,2,1,power_law_1.2,0.4392127990722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,2,1,power_law_1.01,0.16461440324783325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,2,1,power_law_1.2,0.68821120262146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,2,1,power_law_1.2,0.1382143974304199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,2,1,power_law_1.2,0.4563583850860596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,2,1,power_law_1.01,0.17310719490051268
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,2,1,power_law_1.2,0.14334720373153687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,2,1,power_law_1.2,0.49341440200805664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,2,1,power_law_1.01,0.186188805103302
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,2,1,power_law_1.2,0.14744319915771484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,2,1,power_law_1.01,0.20315520763397216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,2,1,power_law_1.2,0.5626880168914795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,2,1,power_law_1.2,0.1506495952606201
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,2,1,power_law_1.01,0.21860480308532715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,2,1,power_law_1.2,0.1561792016029358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,2,balanced,0.6120426654815674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,2,1,power_law_1.01,0.2575167894363403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,2,1,power_law_1.2,1.9460672378540038
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,2,1,power_law_1.2,0.16838400363922118
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,2,1,power_law_1.01,0.23899519443511963
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,2,1,power_law_1.2,0.17244160175323486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,2,1,power_law_1.01,0.2794431924819946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,2,1,power_law_1.01,0.2827967882156372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,2,1,power_law_1.2,2.483731269836426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,2,1,power_law_1.01,0.3355456113815308
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,2,1,power_law_1.01,0.35085439682006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,2,1,power_law_1.2,0.6202688217163086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,2,1,power_law_1.01,0.4218560218811035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,2,1,power_law_1.01,1.058886432647705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,2,1,power_law_1.01,0.5085311889648437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,2,1,power_law_1.2,0.7279295921325684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,2,1,power_law_1.2,0.1885632038116455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,2,1,power_law_1.01,1.2872447967529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,2,1,power_law_1.01,0.6186304092407227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,2,1,power_law_1.2,3.5686336517333985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,2,1,power_law_1.2,0.8525759696960449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,2,1,power_law_1.2,0.7483583927154541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,2,1,power_law_1.2,0.20298879146575927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,2,1,power_law_1.2,0.22486400604248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,2,1,power_law_1.01,0.8426303863525391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,2,1,power_law_1.2,0.7963967800140381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,2,1,power_law_1.01,1.756902313232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,2,1,power_law_1.2,0.2622720003128052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,2,1,power_law_1.2,0.24657919406890869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,2,1,power_law_1.2,0.9131327629089355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,2,1,power_law_1.01,1.2913408279418945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,2,balanced,0.043951998154322304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,2,1,power_law_1.2,0.2868160009384155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,2,balanced,0.0674720009167989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,2,balanced,0.09761599699656169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,2,1,power_law_1.01,2.2361663818359374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,2,1,power_law_1.2,1.044326400756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,2,1,power_law_1.2,0.2928384065628052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,2,1,power_law_1.2,4.643443298339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,2,1,power_law_1.2,0.34358398914337157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,2,balanced,0.054416000843048096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,2,balanced,0.060218666990598045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,2,1,power_law_1.2,1.2950464248657227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,2,balanced,0.09666666388511658
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,2,1,power_law_1.2,0.3854207992553711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,2,balanced,0.13657599687576294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,2,balanced,0.2403786579767863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,2,1,power_law_1.2,1.0888256072998046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,2,balanced,0.43188798427581787
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,2,1,power_law_1.2,0.4337920188903809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,2,1,power_law_1.01,3.1777408599853514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,2,balanced,0.43487465381622314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,2,1,power_law_1.2,1.5569472312927246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,2,balanced,0.43928531805674237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,2,1,power_law_1.2,0.5213759899139404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,2,balanced,0.6129653453826904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,2,1,power_law_1.2,1.31461763381958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,2,balanced,0.6190666755040487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,2,balanced,0.14152000347773233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,2,1,power_law_1.2,0.6364416122436524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,2,balanced,0.7113333543141683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,2,1,power_law_1.2,2.0991872787475585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,2,balanced,0.7151040236155192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,2,1,power_law_1.2,1.7941055297851562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,2,1,power_law_1.2,0.858022403717041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,2,balanced,0.7278133233388265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,2,1,power_law_1.01,4.107852935791016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,2,balanced,0.7396639982859293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,2,power_law_1.01,0.06759679913520814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,2,balanced,0.7481386661529541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,2,1,power_law_1.2,2.71276798248291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,2,1,power_law_1.2,1.369657611846924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,2,power_law_1.01,0.09941120147705078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,2,1,power_law_1.2,2.2533376693725584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,2,balanced,0.027530667682488758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,2,balanced,0.03065599997838338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,2,power_law_1.01,0.1592128038406372
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,2,balanced,0.04891733328501383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,2,balanced,0.06836266815662384
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,2,balanced,0.10514133175214131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,2,balanced,0.2318293253580729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,2,balanced,0.41468266646067303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,2,balanced,0.41818666458129883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,2,balanced,0.4191253185272217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,2,1,power_law_1.2,3.9826942443847657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,2,balanced,0.42156799634297687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,2,balanced,0.4215786854426066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,2,balanced,0.42774399121602374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,2,balanced,0.4664906660715739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,2,balanced,0.7724853356679281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,2,balanced,0.8219573497772217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,2,balanced,0.8696800072987875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,2,balanced,0.1725119948387146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,2,balanced,0.1742080052693685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,2,balanced,0.9599626859029134
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,2,balanced,0.17685333887736002
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,2,balanced,0.1787253419558207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,2,1,power_law_1.2,5.115052795410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,2,balanced,0.18179200092951456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,2,balanced,1.0588266849517822
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,2,balanced,0.18440000216166177
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,2,balanced,0.1881813406944275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,2,balanced,0.19198399782180786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,2,balanced,1.4551040331522624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,2,balanced,0.1983840068181356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,2,balanced,0.19296000401178995
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,2,balanced,0.19696533679962158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,2,balanced,0.4712586800257365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,2,balanced,1.6786452929178874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,2,balanced,0.4867733319600423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,2,balanced,0.4435946544011434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,2,power_law_1.01,0.06108800172805786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,2,balanced,0.49923733870188397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,2,balanced,0.4449919859568278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,2,balanced,0.512725313504537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,2,power_law_1.01,0.0934719979763031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,2,power_law_1.01,0.23311998844146728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,2,balanced,2.361776034037272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,2,balanced,0.5362879832585653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,2,power_law_1.01,0.11388159990310669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,2,power_law_1.01,0.28339200019836425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,2,1,power_law_1.2,3.200377655029297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,2,power_law_1.01,0.170687997341156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,2,power_law_1.01,0.3669183969497681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,2,balanced,3.0214665730794272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,2,power_law_1.01,0.2123392105102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,2,power_law_1.01,0.48581762313842775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,2,balanced,0.2074186603228251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,2,power_law_1.01,0.2836287975311279
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,2,balanced,0.20869867006937662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,2,power_law_1.01,0.4702847957611084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,2,power_law_1.01,0.06105599999427795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,2,power_law_1.01,0.35501439571380616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,2,power_law_1.01,0.5233280181884765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,2,balanced,0.4525066614151001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,2,power_law_1.01,0.3504256010055542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,2,balanced,0.45877333482106525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,2,1,power_law_1.2,4.1463359832763675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,2,power_law_1.01,0.5084415912628174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,2,balanced,0.5964106718699137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,2,power_law_1.01,0.3739840030670166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,2,balanced,0.6471999883651733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,2,power_law_1.01,0.5291520118713379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,2,balanced,0.7619040012359619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,2,balanced,0.8742559750874838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,2,balanced,1.2944693565368652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,2,balanced,0.217631995677948
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,2,balanced,0.23154133558273315
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,2,balanced,0.24385599295298258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,2,balanced,1.5364160537719727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,2,balanced,0.3343040148417155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,2,balanced,4.856341361999512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,2,balanced,0.29815467198689777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,2,balanced,0.4655199845631917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,2,balanced,0.5777279933293661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,2,balanced,0.4784533182779948
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,2,balanced,0.503877321879069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,2,balanced,0.4910186529159546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,2,power_law_1.01,0.37708799839019774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,2,balanced,0.5044906536738077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,2,balanced,0.5299040079116821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,2,power_law_1.01,0.39219839572906495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,2,balanced,0.5794186592102051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,2,balanced,0.6224426825841268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,2,power_law_1.01,0.4072000026702881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,2,balanced,0.7113280296325684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,2,power_law_1.01,0.029023998975753786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,2,power_law_1.01,0.4153855800628662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,2,balanced,6.831050872802734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,2,balanced,0.8225119908650717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,2,power_law_1.01,0.04596480131149292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,2,power_law_1.01,0.43402881622314454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,2,power_law_1.2,0.06768640279769897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,2,balanced,1.1986400286356609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,2,balanced,2.2056053479512534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,2,power_law_1.01,0.4383232116699219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,2,balanced,1.400752067565918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,2,balanced,0.7184693018595377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,2,power_law_1.01,0.4680960178375244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,2,power_law_1.01,0.6300735950469971
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,2,balanced,0.9314986864725748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,2,balanced,2.8759145736694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,2,power_law_1.01,0.5080959796905518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,2,balanced,1.9978346824645996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,2,power_law_1.01,0.6248511791229248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,2,power_law_1.2,0.059673601388931276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,2,power_law_1.01,0.5782144069671631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,2,power_law_1.01,0.6298304080963135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,2,power_law_1.01,0.09409919977188111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,2,balanced,2.6000266075134277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,2,power_law_1.01,0.6539455890655518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,2,power_law_1.01,0.1115007996559143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,2,power_law_1.01,0.6510079860687256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,2,balanced,4.227312088012695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,2,power_law_1.01,0.05580800175666809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,2,power_law_1.01,0.1589184045791626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,2,power_law_1.2,0.10062719583511352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,2,power_law_1.01,0.671020793914795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,2,power_law_1.01,0.7955904006958008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,2,power_law_1.01,0.07443199753761291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,2,power_law_1.01,0.20302081108093262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,2,power_law_1.2,0.16494079828262329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,2,power_law_1.01,0.7239488124847412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,2,balanced,3.7848641077677407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,2,power_law_1.01,0.09591680169105529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,2,power_law_1.01,0.2317888021469116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,2,power_law_1.2,0.21338880062103271
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,2,power_law_1.01,0.1112064003944397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,2,power_law_1.2,0.2649471998214722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,2,balanced,5.57478396097819
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,2,power_law_1.01,0.1450047969818115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,2,power_law_1.2,0.3484544038772583
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,2,power_law_1.01,0.15136640071868895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,2,balanced,4.969690640767415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,2,power_law_1.2,0.4892159938812256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,2,power_law_1.01,0.1605183959007263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,2,power_law_1.01,0.15488640069961548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,2,power_law_1.2,0.4769728183746338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,2,power_law_1.2,0.06051200032234192
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,2,power_law_1.01,0.16338560581207276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,2,power_law_1.01,0.9468031883239746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,2,power_law_1.2,0.09304320216178893
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,2,power_law_1.01,0.16462080478668212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,2,power_law_1.01,0.790015983581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,2,power_law_1.2,0.1068992018699646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,2,power_law_1.01,0.34530560970306395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,2,power_law_1.01,0.17252479791641234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,2,power_law_1.01,1.2552831649780274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,2,power_law_1.2,0.09283199906349182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,2,power_law_1.2,0.15903359651565552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,2,power_law_1.01,0.34272639751434325
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,2,power_law_1.01,0.18343039751052856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,2,power_law_1.01,0.8732671737670898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,2,power_law_1.2,0.10919680595397949
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,2,power_law_1.01,0.17622400522232057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,2,power_law_1.2,0.19248640537261963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,2,power_law_1.01,0.33974399566650393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,2,power_law_1.01,1.0421119689941407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,2,power_law_1.2,0.15268479585647582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,2,power_law_1.01,0.1818112015724182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,2,power_law_1.01,1.5579263687133789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,2,power_law_1.01,0.3472896099090576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,2,power_law_1.2,0.26892800331115724
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,2,power_law_1.01,0.19779839515686035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,2,power_law_1.2,0.19689600467681884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,2,power_law_1.2,0.47739520072937014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,2,power_law_1.01,1.1897600173950196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,2,power_law_1.01,0.20434560775756835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,2,power_law_1.2,0.3552767992019653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,2,power_law_1.2,0.23220479488372803
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,2,power_law_1.01,0.22846720218658448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,2,power_law_1.2,0.37392640113830566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,2,power_law_1.2,0.3154047966003418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,2,power_law_1.01,1.5433856010437013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,2,power_law_1.2,0.37205119132995607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,2,power_law_1.2,0.34344959259033203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,2,power_law_1.2,0.37701759338378904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,2,power_law_1.2,0.3480832099914551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,2,power_law_1.01,1.9212991714477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,2,power_law_1.2,0.3918015956878662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,2,power_law_1.2,0.3505280017852783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,2,power_law_1.01,0.38448638916015626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,2,power_law_1.2,0.37392640113830566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,2,power_law_1.01,0.41187200546264646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,2,power_law_1.2,0.5243135929107666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,2,power_law_1.01,2.599718475341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,2,power_law_1.01,0.4207104206085205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,2,power_law_1.01,0.23666560649871826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,2,power_law_1.2,0.5463744163513183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,2,power_law_1.2,0.02911359965801239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,2,power_law_1.2,0.6092671871185302
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,2,power_law_1.2,0.045363199710845944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,2,power_law_1.2,0.6325056076049804
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,2,power_law_1.2,0.05490559935569763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,2,power_law_1.01,3.3549503326416015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,2,power_law_1.2,0.4082047939300537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,2,power_law_1.2,0.4108543872833252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,2,power_law_1.2,0.40665597915649415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,2,power_law_1.01,2.197216033935547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,2,power_law_1.2,0.4416959762573242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,2,power_law_1.01,0.2719167947769165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,2,power_law_1.2,0.46184959411621096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,2,power_law_1.01,0.33863039016723634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,2,power_law_1.01,4.936665725708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,2,power_law_1.2,0.48236799240112305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,2,power_law_1.01,2.775276756286621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,2,power_law_1.01,0.3845184087753296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,2,power_law_1.2,0.5263872146606445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,2,power_law_1.2,0.07240960001945496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,2,power_law_1.01,0.49799041748046874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,2,power_law_1.2,0.5961023807525635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,2,power_law_1.2,0.09082239866256714
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,2,power_law_1.01,0.6164224147796631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,2,power_law_1.2,0.4162559986114502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,2,power_law_1.01,0.4372608184814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,2,power_law_1.2,0.4438271999359131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,2,power_law_1.01,3.96295051574707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,2,power_law_1.01,0.46002559661865233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,2,power_law_1.2,0.4627903938293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,2,power_law_1.01,6.299046325683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,2,power_law_1.01,0.47813758850097654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,2,1,balanced,0.045519997676213585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,2,1,balanced,0.06222933530807495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,2,1,balanced,0.09705600142478943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,2,1,balanced,0.18004266421000162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,2,power_law_1.01,0.5241343975067139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,2,1,balanced,0.3312000036239624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,2,1,balanced,0.6018186807632446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,2,power_law_1.01,0.610918378829956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,2,1,balanced,0.6032853523890177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,2,power_law_1.2,0.679091215133667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,2,1,balanced,0.6046773195266724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,2,power_law_1.01,0.7059328079223632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,2,power_law_1.01,5.247699356079101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,2,1,balanced,0.606106678644816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,2,power_law_1.2,0.8341888427734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,2,1,balanced,0.6078986724217733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,2,power_law_1.01,0.8661248207092285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,2,1,balanced,0.6096053520838419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,2,1,balanced,0.6161119937896729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,2,power_law_1.2,0.4855936050415039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,2,1,balanced,0.6172106663386027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,2,power_law_1.01,1.0323967933654785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,2,power_law_1.2,0.6647808074951171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,2,1,balanced,0.6243733167648315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,2,1,balanced,0.691759983698527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,2,power_law_1.01,1.4052032470703124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,2,1,balanced,0.6964106559753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,2,power_law_1.2,0.11468160152435303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,2,1,balanced,0.7104907035827637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,2,power_law_1.2,0.13663359880447387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,2,1,balanced,0.7380479971567789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,2,1,balanced,0.03378133227427801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,2,power_law_1.01,1.7278207778930663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,2,1,balanced,0.767520030339559
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,2,power_law_1.2,0.149344003200531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,2,1,balanced,0.8257119655609131
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,2,power_law_1.2,0.14728959798812866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,2,power_law_1.2,0.9872896194458007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,2,1,balanced,0.8933013280232748
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,2,power_law_1.2,0.14789119958877564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,2,power_law_1.2,0.5443327903747559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,2,1,balanced,1.2224640051523845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,2,power_law_1.2,0.1604416012763977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,2,power_law_1.2,1.2576448440551757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,2,power_law_1.2,0.6290559768676758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,2,1,balanced,1.3794186909993489
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,2,power_law_1.01,0.8534784317016602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,2,1,balanced,1.934597333272298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,2,1,balanced,0.05739733576774597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,2,1,balanced,0.08780266841252644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,2,1,balanced,0.13404800494511923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,2,1,balanced,0.23518399397532144
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,2,power_law_1.01,1.5274944305419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,2,1,balanced,0.4251519838968913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,2,1,balanced,2.551647981007894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,2,1,balanced,0.42900268236796063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,2,1,balanced,0.43003201484680176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,2,power_law_1.01,2.4597824096679686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,2,power_law_1.2,0.7125376224517822
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,2,power_law_1.2,0.16317440271377565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,2,power_law_1.2,0.7053760051727295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,2,1,balanced,3.7028586069742837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,2,1,balanced,0.04342400034268697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,2,1,balanced,0.05807466804981232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,2,power_law_1.2,0.740390396118164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,2,power_law_1.01,3.1583295822143556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,2,power_law_1.2,0.8280896186828614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,2,power_law_1.2,0.908358383178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,2,1,balanced,0.4288906653722127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,2,1,balanced,0.43040533860524494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,2,1,balanced,0.433296004931132
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,2,power_law_1.2,0.17093119621276856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,2,power_law_1.2,0.7308991909027099
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,2,power_law_1.2,0.18999040126800537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,2,1,balanced,0.09013866384824117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,2,1,balanced,0.1413706640402476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,2,1,balanced,0.25036799907684326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,2,power_law_1.2,0.1783552050590515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,2,power_law_1.2,0.8968128204345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,2,1,balanced,0.44523199399312335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,2,power_law_1.2,0.18461439609527588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,2,power_law_1.2,1.0684032440185547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,2,1,balanced,5.074010531107585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,2,power_law_1.2,1.0664640426635743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,2,1,balanced,0.43673598766326904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,2,power_law_1.2,1.5872447967529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,2,power_law_1.2,1.2309887886047364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,2,1,balanced,0.45048534870147705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,2,power_law_1.2,2.2426624298095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,2,1,balanced,0.45133864879608154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,2,power_law_1.2,1.5703807830810548
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,2,power_law_1.2,0.21010560989379884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,2,1,balanced,0.4535786708196004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,2,1,balanced,0.4573066631952922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,2,power_law_1.2,0.21590399742126465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,2,1,balanced,0.45947198073069256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,2,power_law_1.2,1.4366463661193847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,2,power_law_1.2,0.23946239948272705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,2,power_law_1.2,1.9438335418701171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,2,power_law_1.01,4.53180160522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,2,power_law_1.2,0.249452805519104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,2,1,balanced,0.029194665451844532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,2,1,balanced,0.03327466547489166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,2,power_law_1.2,0.2842688083648682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,2,power_law_1.2,1.7925504684448241
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,2,power_law_1.2,0.35878400802612304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,2,power_law_1.2,2.6670656204223633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,2,power_law_1.2,0.4183743953704834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,2,power_law_1.2,2.6304384231567384
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,2,power_law_1.2,0.5483903884887695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,2,1,balanced,0.4639146725336711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,2,1,balanced,0.46783467133839923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,2,power_law_1.2,2.806713676452637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,2,power_law_1.2,0.6202432155609131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,2,power_law_1.01,6.08326416015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,2,1,balanced,0.048351998130480446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,2,1,balanced,0.06782933572928111
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,2,power_law_1.2,0.9138879776000977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,2,1,balanced,0.10496532917022705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,2,power_law_1.2,3.3999744415283204
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,2,power_law_1.2,1.564038372039795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,2,power_law_1.2,4.084511947631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,2,1,power_law_1.01,0.0321727991104126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,2,1,balanced,0.43718934059143066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,2,1,power_law_1.01,0.054745602607727054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,2,1,balanced,0.476800004641215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,2,1,balanced,0.4436426560084025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,2,1,balanced,0.4840853214263916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,2,1,balanced,0.4487520058949788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,2,1,power_law_1.01,0.04622719883918762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,2,power_law_1.2,4.784159851074219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,2,1,balanced,0.45872533321380615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,2,1,balanced,0.4726773500442505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,2,1,balanced,0.5092213153839111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,2,1,balanced,0.5455253521601359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,2,1,balanced,0.6255840063095093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,2,1,balanced,0.7036426862080892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,2,power_law_1.2,3.354899215698242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,2,1,balanced,1.0570027033487956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,2,1,balanced,0.49083733558654785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,2,1,balanced,1.2280746301015217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,2,power_law_1.2,6.338105773925781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,2,1,power_law_1.01,0.05991680026054382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,2,1,power_law_1.01,0.09497600197792053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,2,1,balanced,1.7585439682006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,2,1,power_law_1.01,0.17487360239028932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,2,1,power_law_1.01,0.25927679538726806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,2,1,balanced,2.293216069539388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,2,power_law_1.2,5.048108673095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,2,1,power_law_1.01,0.3303168058395386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,2,1,power_law_1.01,0.03782399892807007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,2,1,power_law_1.01,0.09159039855003356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,2,1,balanced,0.5058506727218628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,2,1,balanced,0.17531200249989828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,2,1,power_law_1.01,0.13271679878234863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,2,1,power_law_1.01,0.45355520248413084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,2,1,balanced,0.17675199111302695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,2,1,balanced,0.17934934298197427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,2,1,power_law_1.01,0.1837440013885498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,2,1,power_law_1.01,0.47927680015563967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,2,1,power_law_1.01,0.23832321166992188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,2,1,power_law_1.01,0.486681604385376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,2,1,power_law_1.01,0.31994879245758057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,2,1,power_law_1.01,0.5017792224884033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,2,power_law_1.2,6.439360046386719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,2,1,power_law_1.01,0.328057599067688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,2,1,power_law_1.01,0.5110144138336181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,2,1,power_law_1.01,0.34000000953674314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,2,1,power_law_1.01,0.056908798217773435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,2,1,balanced,3.367274602254232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,2,1,balanced,0.5351786613464355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,2,1,power_law_1.01,0.3559743881225586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,2,1,balanced,0.5588106711705526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,2,1,balanced,0.18115200599034628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,2,1,balanced,0.18284799655278525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,2,1,power_law_1.01,0.3738111972808838
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,2,1,balanced,0.18617600202560425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,2,1,balanced,0.1871839960416158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,2,1,power_law_1.01,0.3985663890838623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,2,1,balanced,4.438410758972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,2,1,power_law_1.2,0.045561599731445315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,2,1,power_law_1.2,0.06037120223045349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,2,1,power_law_1.01,0.5466047763824463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,2,1,power_law_1.2,0.0942143976688385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,2,1,power_law_1.01,0.08718720078468323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,2,1,power_law_1.01,0.5582655906677246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,2,1,power_law_1.2,0.1625920057296753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,2,1,balanced,0.61134934425354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,2,1,power_law_1.01,0.13702399730682374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,2,1,power_law_1.01,0.6202239990234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,2,1,power_law_1.2,0.2333440065383911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,2,1,power_law_1.01,0.20315520763397216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,2,1,power_law_1.01,0.42184958457946775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,2,1,power_law_1.01,0.6300735950469971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,2,1,power_law_1.01,0.2599744081497192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,2,1,power_law_1.01,0.4728447914123535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,2,1,power_law_1.01,0.6217728137969971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,2,1,power_law_1.01,0.3406719923019409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,2,1,power_law_1.01,0.4183807849884033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,2,1,power_law_1.01,0.6670144081115723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,2,1,power_law_1.01,0.4327231884002686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,2,1,power_law_1.01,0.7248447895050049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,2,1,power_law_1.01,0.467577600479126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,2,1,balanced,0.6884160041809082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,2,1,balanced,0.19227200746536255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,2,1,power_law_1.01,0.7688704013824463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,2,1,power_law_1.01,0.026163199543952943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,2,1,balanced,0.19618666172027588
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,2,1,balanced,0.2217866579691569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,2,1,power_law_1.2,0.2959808111190796
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,2,1,balanced,0.2264960010846456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,2,1,power_law_1.01,0.8808575630187988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,2,1,power_law_1.2,0.43909759521484376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,2,1,power_law_1.01,1.0113856315612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,2,1,power_law_1.01,0.3575103998184204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,2,1,power_law_1.2,0.466918420791626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,2,power_law_1.2,5.287404632568359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,2,1,power_law_1.01,0.3675839900970459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,2,1,power_law_1.01,1.2667200088500976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,2,1,power_law_1.2,0.47391362190246583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,2,1,power_law_1.2,0.03308799862861633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,2,1,power_law_1.01,0.378873610496521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,2,1,power_law_1.2,0.05589119791984558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,2,1,power_law_1.2,0.4934400081634521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,2,1,power_law_1.01,0.5247104167938232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,2,1,balanced,1.0275413195292156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,2,1,power_law_1.01,1.5408320426940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,2,1,power_law_1.2,0.5091584205627442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,2,1,power_law_1.01,0.5880640029907227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,2,1,balanced,1.1557119687398274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,2,1,power_law_1.2,0.540825605392456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,2,1,power_law_1.01,0.7161471843719482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,2,1,balanced,1.634463946024577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,2,1,power_law_1.2,0.5772096157073975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,2,1,power_law_1.01,0.8477120399475098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,2,1,power_law_1.2,0.636575984954834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,2,1,balanced,2.1315306027730307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,2,1,power_law_1.01,1.1121151924133301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,2,1,power_law_1.01,0.38572158813476565
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,2,1,power_law_1.01,0.030534398555755616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,2,1,balanced,0.23939200242360434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,2,1,balanced,3.0958452224731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,2,1,power_law_1.01,0.3983488082885742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,2,1,balanced,0.24038400252660116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,2,1,power_law_1.01,0.04714879989624023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,2,1,balanced,0.25783999760945636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,2,1,balanced,0.27746667464574176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,2,1,power_law_1.01,0.40339198112487795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,2,1,power_law_1.01,0.06581760048866273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,2,1,balanced,0.2977919975916545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,2,1,balanced,0.40511465072631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,2,1,power_law_1.01,0.4214784145355225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,2,1,power_law_1.01,0.08572800159454345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,2,1,balanced,0.38151999314626056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,2,1,balanced,4.071152051289876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,2,1,power_law_1.01,0.4381247997283936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,2,1,balanced,0.8314399719238281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,2,1,power_law_1.01,0.11224960088729859
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,2,1,balanced,0.5413866837819418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,2,1,power_law_1.01,0.45765118598937987
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,2,1,power_law_1.01,0.13998719453811645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,2,1,balanced,0.7661440372467041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,2,1,power_law_1.01,0.47164158821105956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,2,1,power_law_1.01,0.14765440225601195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,2,1,balanced,0.9892533620198568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,2,1,power_law_1.01,1.379468822479248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,2,1,power_law_1.01,0.5363200187683106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,2,1,power_law_1.01,2.0794559478759767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,2,1,power_law_1.2,0.03811199963092804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,2,1,power_law_1.01,0.5912896156311035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,2,1,power_law_1.01,1.921228790283203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,2,1,power_law_1.2,0.05723519921302796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,2,1,power_law_1.01,0.7070144176483154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,2,1,power_law_1.2,0.08785920143127442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,2,1,power_law_1.2,0.6392064094543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,2,1,power_law_1.2,0.1324671983718872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,2,1,power_law_1.01,0.8239232063293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,2,1,power_law_1.01,2.627462387084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,2,1,power_law_1.2,0.08659200072288513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,2,1,power_law_1.2,0.6524543762207031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,2,1,power_law_1.2,0.17886719703674317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,2,1,power_law_1.2,0.1320896029472351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,2,1,power_law_1.01,1.0614272117614747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,2,1,power_law_1.2,0.2386240005493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,2,1,power_law_1.01,0.15044480562210083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,2,1,power_law_1.2,0.16306560039520263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,2,1,power_law_1.2,0.336844801902771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,2,1,power_law_1.01,1.295193576812744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,2,1,power_law_1.2,0.2151360034942627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,2,1,power_law_1.01,3.740537643432617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,2,1,power_law_1.2,0.3110719919204712
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,2,1,power_law_1.2,0.02736639976501465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,2,1,power_law_1.01,1.768832015991211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,2,1,power_law_1.2,0.33130879402160646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,2,1,power_law_1.2,0.031148800253868104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,2,1,power_law_1.01,2.453068733215332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,2,1,power_law_1.2,0.04818559885025024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,2,1,power_law_1.2,0.696723222732544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,2,1,power_law_1.01,2.24149112701416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,2,1,power_law_1.01,0.15368959903717042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,2,1,power_law_1.01,5.060883331298828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,2,1,power_law_1.01,0.1605504035949707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,2,1,power_law_1.2,0.3630592107772827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,2,1,power_law_1.01,3.54716796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,2,1,power_law_1.01,0.16749440431594848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,2,1,power_law_1.2,0.3646464109420776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,2,balanced,0.054383998115857445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,2,balanced,0.06841599941253662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,2,1,power_law_1.01,3.181011199951172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,2,balanced,0.094842662413915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,2,balanced,0.17219199736913046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,2,balanced,0.27189866701761883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,2,1,power_law_1.2,0.3428031921386719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,2,balanced,0.3504106601079305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,2,balanced,0.3506133159001668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,2,1,power_law_1.2,0.3565632104873657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,2,1,power_law_1.2,0.06261759996414185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,2,1,power_law_1.2,0.3765120029449463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,2,1,power_law_1.2,0.08108159899711609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,2,1,power_law_1.2,0.7467135906219482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,2,1,power_law_1.01,4.621376037597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,2,1,power_law_1.2,0.40558719635009766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,2,1,power_law_1.2,0.10686719417572021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,2,balanced,0.037647999823093414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,2,balanced,0.05813866853713989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,2,balanced,0.07684266567230225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,2,1,power_law_1.2,0.42605438232421877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,2,1,power_law_1.2,0.13630080223083496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,2,1,power_law_1.01,0.17258880138397217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,2,1,power_law_1.2,0.49453439712524416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,2,1,power_law_1.2,0.1458624005317688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,2,1,power_law_1.2,0.37759358882904054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,2,1,power_law_1.01,0.18442879915237426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,2,1,power_law_1.2,0.42444162368774413
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,2,1,power_law_1.2,0.1515071988105774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,2,1,power_law_1.2,0.38730878829956056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,2,1,power_law_1.01,0.20320000648498535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,2,balanced,0.3545440038045247
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,2,1,power_law_1.2,0.154310405254364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,2,1,power_law_1.2,0.40334720611572267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,2,balanced,0.39900799592336017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,2,1,power_law_1.01,0.21739521026611328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,2,balanced,0.3985653320948283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,2,1,power_law_1.2,0.1617408037185669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,2,1,power_law_1.2,0.40748162269592286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,2,1,power_law_1.01,0.25411200523376465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,2,1,power_law_1.2,0.16787840127944947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,2,1,power_law_1.01,4.114796829223633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,2,1,power_law_1.2,0.4232319831848145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,2,1,power_law_1.01,0.23813118934631347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,2,1,power_law_1.2,0.17626880407333373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,2,balanced,0.12146666646003723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,2,power_law_1.01,0.06177279949188232
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,2,1,power_law_1.01,0.2778815984725952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,2,1,power_law_1.2,0.19189759492874145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,2,power_law_1.01,0.08224639892578126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,2,1,power_law_1.01,0.2811840057373047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,2,1,power_law_1.2,0.20794880390167236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,2,power_law_1.01,0.11388159990310669
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,2,1,power_law_1.2,0.22115840911865234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,2,power_law_1.01,0.1590656042098999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,2,1,power_law_1.2,0.44106240272521974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,2,1,power_law_1.2,0.26284160614013674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,2,power_law_1.01,0.2206399917602539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,2,1,power_law_1.2,0.47679362297058103
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,2,1,power_law_1.2,0.2468735933303833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,2,power_law_1.01,0.2863424062728882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,2,1,power_law_1.2,0.549177598953247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,2,power_law_1.01,0.30600318908691404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,2,1,power_law_1.2,0.8073216438293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,2,power_law_1.01,0.3079360008239746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,2,1,power_law_1.2,0.4407936096191406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,2,balanced,0.18556267023086548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,2,balanced,0.2411200006802877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,2,power_law_1.01,0.3666048049926758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,2,balanced,0.24440000454584757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,2,1,power_law_1.2,0.9152959823608399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,2,balanced,0.24551467100779215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,2,power_law_1.01,0.37920639514923093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,2,1,power_law_1.01,0.33393919467926025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,2,power_law_1.01,0.38569600582122804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,2,1,power_law_1.2,1.0292096138000488
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,2,1,power_law_1.01,0.35032958984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,2,power_law_1.01,0.39565439224243165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,2,balanced,0.40195198853810626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,2,1,power_law_1.01,0.4225471973419189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,2,power_law_1.01,0.3800895929336548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,2,1,power_law_1.2,1.3062335968017578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,2,1,power_law_1.2,0.2878783941268921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,2,power_law_1.01,0.4108543872833252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,2,1,power_law_1.01,0.5005760192871094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,2,1,power_law_1.2,0.2937664031982422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,2,1,power_law_1.2,0.6103487968444824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,2,power_law_1.01,0.41292800903320315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,2,1,power_law_1.2,1.5603520393371582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,2,1,power_law_1.2,0.36375041007995607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,2,1,power_law_1.2,0.45728001594543455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,2,power_law_1.01,0.4298431873321533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,2,1,power_law_1.2,0.7302720069885253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,2,balanced,0.26834134260813397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,2,1,power_law_1.2,0.36487040519714353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,2,balanced,0.2691626747449239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,2,power_law_1.01,0.4463935852050781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,2,balanced,0.2720906734466553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,2,balanced,0.27591999371846515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,2,1,power_law_1.2,0.46634879112243655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,2,1,power_law_1.2,0.8645824432373047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,2,balanced,0.2793546716372172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,2,power_law_1.01,0.5145919799804688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,2,balanced,0.28818132479985553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,2,balanced,0.2981333335240682
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,2,1,power_law_1.2,0.5132927894592285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,2,power_law_1.01,0.5687488079071045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,2,balanced,0.40491199493408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,2,1,power_law_1.2,1.1357248306274415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,2,balanced,0.40828800201416016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,2,1,power_law_1.2,0.6314752101898193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,2,1,power_law_1.2,1.4049344062805176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,2,1,power_law_1.01,0.611891222000122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,2,1,power_law_1.2,0.8569600105285644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,2,1,power_law_1.2,0.49592318534851076
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,2,1,power_law_1.01,0.8416255950927735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,2,1,power_law_1.2,0.5651072025299072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,2,1,power_law_1.2,1.9360128402709962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,2,balanced,0.3075253367424011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,2,balanced,0.3298719922701518
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,2,1,power_law_1.01,1.3025919914245605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,2,balanced,0.3729493220647176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,2,power_law_1.01,0.683289623260498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,2,balanced,0.4140213330586751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,2,1,power_law_1.2,2.4772415161132812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,2,balanced,0.6133706569671631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,2,power_law_1.01,0.8594047546386718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,2,balanced,0.6968479951222738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,2,balanced,0.9951146443684896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,2,power_law_1.01,1.1140416145324707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,2,1,power_law_1.2,1.3637439727783203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,2,balanced,1.1768373648325603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,2,power_law_1.2,0.06247040033340454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,2,1,power_law_1.2,3.558726501464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,2,power_law_1.01,1.4478591918945312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,2,power_law_1.2,0.077183997631073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,2,balanced,0.41915734608968097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,2,power_law_1.2,0.10311039686203002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,2,balanced,0.4273759921391805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,2,power_law_1.01,0.05418239831924439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,2,power_law_1.2,0.16039040088653564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,2,power_law_1.01,1.9251264572143554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,2,power_law_1.01,0.07401599884033203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,2,power_law_1.01,0.09304959774017334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,2,power_law_1.01,0.11769599914550781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,2,1,power_law_1.2,2.1144128799438477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,2,1,power_law_1.2,4.656441497802734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,2,power_law_1.01,0.1431615948677063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,2,power_law_1.2,0.05260159969329834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,2,power_law_1.01,0.19225599765777587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,2,balanced,1.7664373715718586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,2,power_law_1.2,0.06818559765815735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,2,power_law_1.01,0.1949504017829895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,2,power_law_1.2,0.09168000221252441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,2,power_law_1.01,0.21342720985412597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,2,1,power_law_1.2,2.642553520202637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,2,balanced,0.4313066800435384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,2,power_law_1.2,0.10634880065917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,2,power_law_1.01,0.23073279857635498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,2,balanced,0.4464000066121419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,2,power_law_1.2,0.16947840452194213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,2,balanced,0.4761439959208171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,2,power_law_1.01,0.24722559452056886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,2,power_law_1.2,0.2812096118927002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,2,power_law_1.01,0.2687616109848022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,2,power_law_1.2,0.30183680057525636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,2,power_law_1.01,0.2694591999053955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,2,power_law_1.01,2.5502464294433596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,2,power_law_1.2,0.3100672006607056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,2,power_law_1.01,0.26389760971069337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,2,1,power_law_1.2,0.6186816215515136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,2,1,power_law_1.2,3.7970176696777345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,2,power_law_1.2,0.3540287971496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,2,power_law_1.01,0.2863744020462036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,2,power_law_1.2,0.36110079288482666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,2,power_law_1.01,0.29726719856262207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,2,1,power_law_1.2,0.7248511791229248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,2,power_law_1.01,0.3090751886367798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,2,balanced,2.255093256632487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,2,power_law_1.01,0.3547391891479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,2,1,power_law_1.2,0.8374143600463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,2,power_law_1.01,3.8080448150634765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,2,balanced,0.5099413394927979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,2,power_law_1.01,0.42926721572875975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,2,balanced,0.6972053050994873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,2,1,power_law_1.2,1.0912960052490235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,2,balanced,3.331808090209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,2,balanced,0.7739840348561605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,2,1,power_law_1.2,4.997638320922851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,2,1,balanced,0.0435146689414978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,2,1,balanced,0.05637866755326589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,2,1,power_law_1.2,1.3178624153137206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,2,1,balanced,0.08691733082135518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,2,1,balanced,0.1620746652285258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,2,1,balanced,0.2725493311882019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,2,1,balanced,0.348416010538737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,2,power_law_1.2,0.38087680339813235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,2,power_law_1.01,4.902016067504883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,2,balanced,4.410671869913737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,2,1,power_law_1.2,1.7928192138671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,2,power_law_1.2,0.1383679986000061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,2,1,power_law_1.01,0.03720319867134094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,2,power_law_1.2,0.19808640480041503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,2,1,power_law_1.01,0.051583999395370485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,2,power_law_1.2,0.20670080184936523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,2,power_law_1.01,0.48623361587524416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,2,power_law_1.2,0.2139904022216797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,2,1,power_law_1.2,2.260895919799805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,2,power_law_1.01,0.6295423984527588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,2,power_law_1.2,0.23873279094696045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,2,power_law_1.2,0.24686720371246337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,2,power_law_1.01,0.7592576026916504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,2,power_law_1.2,0.2616384029388428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,2,1,balanced,0.34700266520182294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,2,power_law_1.2,0.269267201423645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,2,1,balanced,0.3478399912516276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,2,power_law_1.2,0.3952255964279175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,2,1,balanced,0.3487199942270915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,2,power_law_1.2,0.272492790222168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,2,1,balanced,0.3507466713587443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,2,power_law_1.2,0.39838719367980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,2,1,balanced,0.3537973165512085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,2,power_law_1.2,0.2854720115661621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,2,1,balanced,0.3888373374938965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,2,1,power_law_1.01,0.08184319734573364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,2,1,balanced,0.39246400197347003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,2,1,balanced,0.3988906542460124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,2,balanced,1.0776426792144775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,2,1,balanced,0.40355201562245685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,2,1,balanced,0.4102026621500651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,2,1,balanced,0.42109866937001544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,2,1,balanced,0.43942399819691974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,2,1,balanced,0.46382399400075275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,2,1,balanced,0.031152000029881794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,2,1,balanced,0.6399306853612264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,2,1,balanced,0.047914668917655945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,2,1,power_law_1.2,3.210732650756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,2,1,balanced,0.0701279987891515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,2,1,balanced,0.11342400312423706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,2,1,balanced,0.18894400199254355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,2,1,balanced,0.24549333254496256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,2,1,balanced,0.2448319991429647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,2,power_law_1.2,0.4063231945037842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,2,1,balanced,0.2479733427365621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,2,power_law_1.2,0.29679360389709475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,2,1,balanced,0.24830933411916098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,2,1,power_law_1.01,0.1478271961212158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,2,1,balanced,0.25031999746958417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,2,power_law_1.2,0.313100790977478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,2,1,balanced,0.2516746719678243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,2,1,power_law_1.01,0.16743680238723754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,2,1,balanced,0.2630133430163066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,2,power_law_1.2,0.35148160457611083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,2,1,power_law_1.01,0.27301759719848634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,2,1,power_law_1.2,4.142367935180664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,2,power_law_1.2,0.42088961601257324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,2,1,power_law_1.01,0.2918463945388794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,2,1,power_law_1.01,0.030265599489212036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,2,power_law_1.01,1.0720959663391114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,2,power_law_1.2,0.48418560028076174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,2,1,power_law_1.01,0.0477183997631073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,2,1,balanced,0.6833120187123617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,2,1,power_law_1.01,0.0700160026550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,2,power_law_1.2,0.6254208087921143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,2,1,balanced,0.9462506771087646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,2,power_law_1.01,1.337504005432129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,2,1,power_law_1.01,0.10433919429779052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,2,1,balanced,1.0651893615722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,2,1,power_law_1.01,0.1304703950881958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,2,1,power_law_1.01,0.19426560401916504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,2,1,balanced,1.5889172554016113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,2,1,balanced,0.2655679980913798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,2,1,power_law_1.01,0.20833280086517333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,2,1,balanced,0.2743413249651591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,2,1,balanced,0.2786879936854045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,2,1,power_law_1.01,0.21038079261779785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,2,1,balanced,0.2844906648000081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,2,1,balanced,1.9764159520467122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,2,1,balanced,0.29632000128428143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,2,1,power_law_1.01,0.21800320148468016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,2,1,power_law_1.01,0.30383360385894775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,2,1,balanced,0.3244746724764506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,2,1,power_law_1.01,0.22780799865722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,2,1,power_law_1.01,0.3061631917953491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,2,1,power_law_1.01,0.2443455934524536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,2,1,power_law_1.01,0.3093055963516235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,2,1,balanced,2.90122127532959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,2,1,power_law_1.01,0.2491391897201538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,2,power_law_1.2,0.774841594696045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,2,power_law_1.2,0.4197824001312256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,2,1,power_law_1.01,0.2573951959609985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,2,power_law_1.2,0.4373119831085205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,2,1,power_law_1.01,0.27073919773101807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,2,balanced,1.2475519975026448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,2,power_law_1.01,1.9165760040283204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,2,1,power_law_1.01,0.27616000175476074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,2,1,balanced,3.919637362162272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,2,1,power_law_1.01,0.29034240245819093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,2,balanced,1.843605359395345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,2,1,power_law_1.2,0.03914240002632141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,2,1,power_law_1.01,0.31423358917236327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,2,1,power_law_1.2,0.05242239832878113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,2,1,balanced,0.3545173406600952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,2,power_law_1.01,2.506284713745117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,2,1,power_law_1.01,0.3642688035964966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,2,1,balanced,0.5471359888712565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,2,balanced,2.3184000651041665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,2,1,balanced,0.597269336382548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,2,1,power_law_1.01,0.4253056049346924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,2,1,power_law_1.01,0.3281536102294922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,2,1,balanced,0.8526079654693604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,2,1,power_law_1.01,0.5272192001342774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,2,1,power_law_1.01,0.3660736083984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,2,1,balanced,0.9860693613688151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,2,power_law_1.2,1.0500479698181153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,2,balanced,3.620965321858724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,2,1,power_law_1.01,0.6346303939819335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,2,1,power_law_1.01,0.3732991933822632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,2,power_law_1.2,0.47390079498291016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,2,power_law_1.01,3.5024959564208986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,2,1,balanced,1.4876906077067058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,2,1,power_law_1.01,0.3852927923202515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,2,1,power_law_1.01,0.8598464012145997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,2,power_law_1.2,0.5222911834716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,2,power_law_1.2,1.3678272247314454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,2,1,balanced,1.8821333249409993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,2,power_law_1.2,0.5850240230560303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,2,1,power_law_1.2,0.08154240250587463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,2,1,power_law_1.01,1.0834879875183105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,2,1,power_law_1.2,0.12836480140686035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,2,power_law_1.2,0.725657606124878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,2,balanced,5.2170718510945635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,2,1,power_law_1.2,0.16301440000534057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,2,1,balanced,2.7795626322428384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,2,1,power_law_1.2,0.030425599217414855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,2,1,power_law_1.01,1.5372096061706544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,2,power_law_1.2,0.8475008010864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,2,1,power_law_1.2,0.26753280162811277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,2,1,power_law_1.2,0.04575360119342804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,2,power_law_1.01,4.924825668334961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,2,1,power_law_1.2,0.07022719979286193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,2,power_law_1.2,1.128870391845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,2,1,power_law_1.2,0.09904000163078308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,2,1,power_law_1.01,1.9822656631469726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,2,1,balanced,3.6791200637817383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,2,1,power_law_1.01,0.39631359577178954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,2,power_law_1.2,2.113350486755371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,2,power_law_1.2,1.432863998413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,2,balanced,0.12566399574279785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,2,balanced,0.20174400011698404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,2,balanced,0.34753600756327313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,2,1,power_law_1.01,2.8927743911743162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,2,balanced,0.6495466629664103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,2,1,power_law_1.2,0.2859776020050049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,2,balanced,1.2088053226470947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,2,1,power_law_1.2,0.3014336109161377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,2,balanced,0.16879467169443765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,2,1,power_law_1.2,0.12112640142440796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,2,1,power_law_1.01,0.40061440467834475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,2,1,power_law_1.01,3.796223831176758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,2,1,power_law_1.01,0.4211775779724121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,2,1,power_law_1.01,0.4703360080718994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,2,1,power_law_1.01,0.51146240234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,2,1,power_law_1.01,0.6120831966400146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,2,power_law_1.2,2.086879920959473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,2,1,power_law_1.2,0.30691840648651125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,2,1,power_law_1.01,0.7175615787506103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,2,balanced,0.3084320028622945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,2,1,power_law_1.2,0.32206718921661376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,2,balanced,0.530346671740214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,2,1,power_law_1.2,0.19265279769897461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,2,1,power_law_1.2,0.33435521125793455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,2,1,power_law_1.2,0.20790400505065917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,2,balanced,0.9984266757965088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,2,power_law_1.2,2.573971176147461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,2,1,power_law_1.2,0.37188479900360105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,2,1,power_law_1.2,0.2161792039871216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,2,1,power_law_1.2,0.22664320468902588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,2,balanced,1.8589386940002441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,2,1,power_law_1.2,0.23471360206604003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,2,1,power_law_1.2,0.25419518947601316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,2,balanced,2.681701342264811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,2,1,power_law_1.2,0.25753600597381593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,2,balanced,1.7483572959899902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,2,power_law_1.2,3.879974365234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,2,1,power_law_1.2,0.2624000072479248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,2,balanced,1.7606666882832844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,2,1,power_law_1.01,0.9439680099487304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,2,balanced,2.749349276224772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,2,balanced,1.7709120114644368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,2,1,power_law_1.01,1.1679936408996583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,2,1,power_law_1.2,0.37925760746002196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,2,balanced,2.757253328959147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,2,1,power_law_1.2,0.3874111890792847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,2,balanced,1.7790452639261882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,2,power_law_1.2,2.7542463302612306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,2,1,power_law_1.2,0.39866878986358645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,2,1,power_law_1.01,1.6370559692382813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,2,balanced,1.7905492782592773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,2,1,power_law_1.2,0.404863977432251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,2,balanced,2.7278079986572266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,2,1,power_law_1.2,0.44047999382019043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,2,1,power_law_1.2,0.27423360347747805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,2,balanced,1.8603040377298992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,2,1,power_law_1.01,2.09747200012207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,2,1,power_law_1.2,0.47531518936157224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,2,1,power_law_1.2,0.2857856035232544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,2,balanced,2.7547785441080728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,2,1,power_law_1.2,0.2959424018859863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,2,balanced,1.8502666155497234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,2,1,power_law_1.2,0.32655360698699953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,2,balanced,2.8080320358276367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,2,1,power_law_1.2,0.37801599502563477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,2,1,power_law_1.01,3.0358144760131838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,2,1,power_law_1.2,0.4292287826538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,2,1,power_law_1.2,0.5410751819610595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,2,balanced,2.817381223042806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,2,balanced,0.16548267006874084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,2,balanced,0.20549867550532022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,2,1,power_law_1.2,0.6442111968994141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,2,balanced,0.3417280117670695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,2,balanced,0.6356853246688843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,2,1,power_law_1.2,0.5272128105163574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,2,balanced,2.8437172571818032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,2,1,power_law_1.01,3.982380676269531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,2,1,power_law_1.2,0.8685824394226074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,2,balanced,1.193658669789632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,2,1,power_law_1.2,0.6241151809692382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,2,balanced,1.8784319559733074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,2,balanced,1.7282400131225586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,2,balanced,2.8989814122517905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,2,power_law_1.01,0.29317760467529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,2,1,power_law_1.2,1.0937600135803223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,2,balanced,1.8930346171061199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,2,balanced,1.7433013916015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,2,power_law_1.01,0.5102272033691406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,2,balanced,2.332538604736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,2,balanced,3.302192052205404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,2,1,power_law_1.2,1.5489727973937988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,2,power_law_1.01,0.6714176177978516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,2,balanced,2.4116479555765786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,2,balanced,3.409754753112793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,2,1,power_law_1.2,0.7373824119567871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,2,power_law_1.01,1.0656703948974608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,2,power_law_1.2,3.767782211303711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,2,balanced,2.5509279568990073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,2,1,power_law_1.2,0.9578944206237793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,2,balanced,3.475269317626953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,2,power_law_1.01,1.4136575698852538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,2,balanced,2.8021653493245444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,2,1,power_law_1.2,1.185542392730713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,2,1,power_law_1.2,1.653971290588379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,2,balanced,3.71836789449056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,2,balanced,3.0492960611979165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,2,power_law_1.01,2.0339136123657227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,2,power_law_1.2,4.854707336425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,2,1,power_law_1.2,2.001055908203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,2,1,power_law_1.2,2.131539154052734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,2,balanced,3.966224034627279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,2,power_law_1.01,2.257151985168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,2,balanced,1.7586453755696614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,2,1,power_law_1.2,2.9125375747680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,2,balanced,1.7738986015319824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,2,power_law_1.01,2.272211265563965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,2,balanced,1.7892319361368816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,2,balanced,5.769856135050456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,2,balanced,1.8171733220418294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,2,1,power_law_1.2,3.820076751708984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,2,power_law_1.01,2.3308095932006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,2,power_law_1.01,0.19431040287017823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,2,balanced,1.847482681274414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,2,power_law_1.01,0.33724799156188967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,2,balanced,1.879242738087972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,2,power_law_1.01,2.3409408569335937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,2,balanced,6.0022023518880205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,2,balanced,1.9327306747436523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,2,power_law_1.01,0.4344831943511963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,2,balanced,1.9817387262980144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,2,balanced,4.749397277832031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,2,power_law_1.01,0.6648704051971436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,2,power_law_1.2,5.164102554321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,2,balanced,2.0362399419148765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,2,power_law_1.01,0.9128895759582519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,2,balanced,2.1284213066101074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,2,power_law_1.2,0.2934720039367676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,2,balanced,9.445274353027344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,2,balanced,5.218976020812988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,2,power_law_1.01,1.362771224975586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,2,balanced,2.3006720542907715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,2,power_law_1.2,0.43671040534973143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,2,power_law_1.01,2.5666879653930663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,2,power_law_1.01,0.19854719638824464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,2,power_law_1.2,0.6568064212799072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,2,balanced,2.4770612716674805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,2,power_law_1.01,0.3311808109283447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,2,power_law_1.01,2.6914047241210937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,2,power_law_1.2,0.9515839576721191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,2,power_law_1.01,0.40025601387023924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,2,balanced,7.509189605712891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,2,balanced,3.7887519200642905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,2,power_law_1.01,0.6461184024810791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,2,power_law_1.2,1.262662410736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,2,power_law_1.01,2.9458560943603516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,2,balanced,11.324843088785807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,2,power_law_1.01,1.4523520469665527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,2,power_law_1.01,0.9359231948852539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,2,balanced,4.161199887593587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,2,power_law_1.2,2.103455924987793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,2,power_law_1.01,1.4972479820251465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,2,power_law_1.01,1.3744192123413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,2,power_law_1.01,3.3043006896972655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,2,balanced,8.649920145670572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,2,power_law_1.01,1.540351963043213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,2,power_law_1.01,1.4713983535766602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,2,balanced,5.937344233194987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,2,power_law_1.01,3.0702592849731447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,2,power_law_1.01,1.6282623291015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,2,power_law_1.01,1.5221952438354491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,2,1,power_law_1.2,3.047558403015137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,2,power_law_1.01,1.7437248229980469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,2,power_law_1.01,1.5867839813232423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,2,power_law_1.01,3.218527984619141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,2,balanced,6.666101455688477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,2,power_law_1.2,2.291814422607422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,2,power_law_1.01,1.8990976333618164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,2,power_law_1.01,1.6170431137084962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,2,balanced,18.075903574625652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,2,1,power_law_1.2,4.091148757934571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,2,power_law_1.2,2.2453887939453123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,2,power_law_1.01,3.420044708251953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,2,balanced,14.787605285644531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,2,power_law_1.2,2.264467239379883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,2,power_law_1.2,0.1972864031791687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,2,power_law_1.01,3.793791961669922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,2,power_law_1.2,0.29205760955810545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,2,balanced,10.226624170939127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,2,power_law_1.2,2.4283519744873048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,2,power_law_1.2,0.4407680034637451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,2,power_law_1.01,2.0886207580566407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,2,power_law_1.2,0.6156544208526611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,2,power_law_1.2,2.661952018737793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,2,power_law_1.01,4.380908966064453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,2,power_law_1.01,2.3567808151245115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,2,power_law_1.2,0.7903103828430176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,2,power_law_1.2,2.8509183883666993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,2,power_law_1.2,1.3240447998046876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,2,power_law_1.01,1.703660774230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,2,balanced,12.701338450113932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,2,power_law_1.01,5.44029426574707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,2,power_law_1.01,1.7273664474487305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,2,balanced,19.927616119384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,2,power_law_1.2,3.089708709716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,2,balanced,23.433865865071613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,2,power_law_1.01,1.7588607788085937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,2,power_law_1.01,2.2197824478149415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,2,power_law_1.2,3.440275192260742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,2,power_law_1.01,1.8444032669067383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,2,power_law_1.2,1.4705984115600585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,2,power_law_1.01,6.338899230957031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,2,power_law_1.01,2.3847103118896484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,2,power_law_1.01,1.9182144165039063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,2,power_law_1.2,1.4442943572998046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,2,power_law_1.2,3.1886335372924806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,2,power_law_1.01,2.6284223556518556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,2,power_law_1.2,1.5706239700317384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,2,power_law_1.01,2.058393669128418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,2,power_law_1.2,1.678656005859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,2,power_law_1.2,3.283039855957031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,2,power_law_1.01,3.0696319580078124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,2,power_law_1.01,2.2255104064941404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,2,balanced,18.770533243815105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,2,power_law_1.01,8.331206512451171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,2,power_law_1.01,2.587808036804199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,2,power_law_1.01,3.5423423767089846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,2,power_law_1.2,3.5570240020751953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,2,power_law_1.01,2.978950309753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,2,power_law_1.2,1.8054655075073243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,2,power_law_1.2,3.896748733520508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,2,power_law_1.01,4.474182510375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,2,power_law_1.2,1.9618047714233398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,2,balanced,30.96515655517578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,2,power_law_1.01,3.7046913146972655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,2,power_law_1.01,10.397695922851563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,2,power_law_1.2,2.182822418212891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,2,power_law_1.2,4.532665634155274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,2,power_law_1.01,5.467654418945313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,2,balanced,35.26981862386068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,2,power_law_1.2,2.4897600173950196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,2,power_law_1.2,5.426105499267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,2,power_law_1.2,2.3630847930908203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,2,balanced,23.730545043945312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,2,power_law_1.01,7.3681793212890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,2,power_law_1.01,4.493529510498047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,2,power_law_1.2,2.487411117553711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,2,power_law_1.2,6.556192016601562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,2,power_law_1.2,2.7171968460083007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,2,power_law_1.01,14.940684509277343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,2,power_law_1.01,5.94207992553711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,2,power_law_1.2,3.156217575073242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,2,power_law_1.01,9.094252777099609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,2,power_law_1.2,0.1972607970237732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,2,power_law_1.2,0.29066879749298097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,2,power_law_1.2,3.7342529296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,2,power_law_1.2,8.820588684082031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,2,power_law_1.2,0.431660795211792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,2,power_law_1.01,7.490470123291016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,2,power_law_1.2,0.6204415798187256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,2,power_law_1.2,4.5784961700439455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,2,power_law_1.2,0.7686975955963135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,2,balanced,40.27102406819662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,2,power_law_1.2,1.4010496139526367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,2,power_law_1.01,12.995820617675781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,2,power_law_1.2,1.4438655853271485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,2,power_law_1.2,5.575551986694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,2,power_law_1.01,19.851239013671876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,2,power_law_1.2,11.331168365478515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,2,power_law_1.2,1.5049663543701173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,2,power_law_1.01,10.702207946777344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,2,power_law_1.2,1.5644351959228515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,2,balanced,46.65931701660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,2,power_law_1.2,1.5875264167785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,2,1,balanced,0.11819199721018474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,2,1,balanced,0.18784532944361368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,2,power_law_1.2,7.586841583251953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,2,1,balanced,0.33641600608825684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,2,power_law_1.2,1.6569599151611327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,2,1,balanced,0.6383413473765055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,2,1,balanced,1.2140586376190186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,2,power_law_1.2,1.7359743118286133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,2,1,balanced,1.7729867299397786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,2,1,balanced,1.7774933179219563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,2,power_law_1.2,1.8044544219970704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,2,1,balanced,1.781925360361735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,2,power_law_1.2,1.900556755065918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,2,1,balanced,1.7874239285786946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,2,power_law_1.01,17.287507629394533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,2,power_law_1.01,13.510765075683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,2,power_law_1.2,16.378553771972655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,2,1,balanced,1.8296906153361003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,2,power_law_1.2,2.0164863586425783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,2,1,balanced,1.8127093315124512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,2,1,balanced,1.8185920715332031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,2,power_law_1.2,2.1089792251586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,2,1,balanced,1.832549254099528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,2,1,balanced,0.13659733533859253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,2,power_law_1.2,9.613728332519532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,2,1,balanced,1.9065279960632324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,2,power_law_1.2,2.2698240280151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,2,1,balanced,1.888821283976237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,2,power_law_1.01,28.195263671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,2,1,balanced,2.700943946838379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,2,power_law_1.2,2.5884607315063475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,2,1,balanced,3.5454025268554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,2,power_law_1.2,3.060838317871094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,2,1,balanced,2.5175253550211587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,2,1,balanced,2.7185866038004556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,2,power_law_1.2,3.78600959777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,2,power_law_1.01,20.418170166015624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,2,power_law_1.2,20.38251495361328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,2,power_law_1.2,13.348287963867188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,2,power_law_1.01,24.256536865234374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,2,1,balanced,4.32484785715739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,2,power_law_1.2,4.497932815551758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,2,1,balanced,4.668282508850098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,2,1,balanced,0.2464746634165446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,2,power_law_1.2,6.224275207519531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,2,1,balanced,6.72593625386556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,2,1,balanced,7.490506490071614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,2,power_law_1.2,7.599276733398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,2,power_law_1.01,25.766033935546876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,2,1,balanced,11.541824340820312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,2,power_law_1.01,40.23894348144531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,2,power_law_1.2,10.943500518798828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,2,power_law_1.2,17.951213073730468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,2,1,power_law_1.01,0.11791360378265381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,2,1,power_law_1.01,0.18655999898910522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,2,power_law_1.01,32.32748718261719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,2,1,power_law_1.01,0.33420159816741946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,2,1,power_law_1.2,0.13509119749069215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,2,1,balanced,0.4503573179244995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,2,1,power_law_1.01,0.5718272209167481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,2,1,balanced,14.311861673990885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,2,1,power_law_1.2,0.24714241027832032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,2,1,power_law_1.01,0.806060791015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,2,1,power_law_1.2,0.4503488063812256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,2,1,power_law_1.01,1.3168576240539551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,2,1,power_law_1.2,0.7085311889648438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,2,1,power_law_1.01,1.4110143661499024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,2,1,power_law_1.2,0.9763392448425293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,2,power_law_1.2,14.132710266113282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,2,1,power_law_1.01,1.4731200218200684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,2,power_law_1.2,30.516876220703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,2,1,power_law_1.2,1.7578943252563477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,2,1,power_law_1.01,1.5312128067016602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,2,1,power_law_1.2,1.9061504364013673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,2,1,power_law_1.01,1.5773247718811034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,2,1,power_law_1.01,1.7371391296386718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,2,1,balanced,21.622873942057293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,2,1,power_law_1.01,1.85479679107666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,2,1,power_law_1.01,2.025312042236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,2,1,power_law_1.2,1.998918342590332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,2,1,power_law_1.01,0.13628159761428832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,2,1,power_law_1.01,2.308051109313965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,2,1,power_law_1.2,2.065363121032715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,2,1,power_law_1.01,2.6027904510498048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,2,1,power_law_1.2,2.0912960052490233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,2,1,balanced,0.8575572967529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,2,1,power_law_1.01,2.908185577392578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,2,1,power_law_1.2,2.2185216903686524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,2,1,power_law_1.2,2.2912576675415037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,2,1,power_law_1.01,3.507769775390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,2,1,power_law_1.2,2.3791744232177736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,2,1,power_law_1.01,2.773459243774414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,2,1,balanced,27.877593994140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,2,1,power_law_1.2,2.4818559646606446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,2,1,power_law_1.01,3.164793586730957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,2,1,power_law_1.2,0.11644159555435181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,2,1,power_law_1.2,0.18563200235366822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,2,1,power_law_1.2,2.6624256134033204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,2,1,power_law_1.2,0.3345855951309204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,2,1,power_law_1.01,3.968697738647461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,2,1,power_law_1.2,0.5268352031707764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,2,1,power_law_1.01,0.2422271966934204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,2,1,power_law_1.2,2.807481575012207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,2,power_law_1.2,20.352467346191407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,2,1,power_law_1.2,0.7274943828582764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,2,1,power_law_1.2,1.3094911575317383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,2,1,power_law_1.2,3.0073471069335938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,2,1,power_law_1.01,4.743443298339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,2,1,power_law_1.2,1.395359992980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,2,1,power_law_1.2,3.120217514038086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,2,1,power_law_1.2,1.4514176368713378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,2,1,power_law_1.2,1.5662400245666503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,2,1,power_law_1.01,6.3874366760253904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,2,1,power_law_1.2,3.3853118896484373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,2,1,power_law_1.2,1.6364608764648438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,2,1,balanced,1.6352267265319824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,2,1,power_law_1.2,1.7677248001098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,2,1,power_law_1.2,3.8102336883544923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,2,1,power_law_1.2,1.9307327270507812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,2,1,power_law_1.01,8.003731536865235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,2,1,power_law_1.2,2.0886016845703126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,2,1,power_law_1.2,4.208844757080078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,2,1,balanced,2.395930608113607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,2,1,power_law_1.2,2.386892890930176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,2,1,balanced,2.4100586573282876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,2,power_law_1.2,25.476882934570312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,2,1,balanced,2.4355252583821616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,2,1,power_law_1.2,2.725651168823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,2,power_law_1.2,43.656961059570314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,2,1,balanced,2.4521867434183755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,4,balanced,0.1627946694691976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,4,balanced,0.16529599825541177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,4,balanced,0.3540266752243042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,2,1,power_law_1.01,0.45241599082946776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,4,balanced,0.3521653413772583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,2,1,power_law_1.01,11.228864288330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,2,1,balanced,2.4426933924357095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,4,balanced,0.350597341855367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,4,balanced,0.35467199484507245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,4,balanced,0.3567146857579549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,4,balanced,0.3567253351211548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,4,balanced,0.3617386817932129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,2,1,power_law_1.2,5.216767883300781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,4,balanced,0.36467734972635907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,4,balanced,0.3691519896189372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,4,balanced,0.3744639952977498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,4,balanced,0.38178666432698566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,4,balanced,0.396565318107605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,4,balanced,0.4089440107345581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,4,balanced,0.4258879820505778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,2,1,power_law_1.2,2.987289619445801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,4,balanced,0.42291732629140216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,4,balanced,0.4982080062230428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,4,balanced,0.5906826655069987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,4,balanced,0.8486773173014323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,2,1,power_law_1.2,6.079731369018555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,4,balanced,1.0224800109863281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,2,1,power_law_1.2,3.6078399658203124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,4,balanced,1.5942293802897136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,4,balanced,2.218597412109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,2,1,power_law_1.2,2.852115249633789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,2,1,power_law_1.01,14.532063293457032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,4,balanced,3.5722134908040366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,2,1,power_law_1.2,8.149440002441406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,2,power_law_1.2,26.415966796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,4,balanced,0.1452959974606832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,4,balanced,0.146506667137146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,4,balanced,0.2683359980583191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,4,balanced,5.311039924621582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,4,balanced,0.26665065685908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,4,balanced,0.2663946747779846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,4,balanced,0.2734559973080953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,4,balanced,0.2765760024388631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,4,balanced,0.27768532435099286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,4,balanced,0.2820853392283122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,4,balanced,0.2817280093828837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,4,balanced,0.2898826599121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,4,balanced,0.2988266746203105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,4,balanced,0.31142934163411456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,2,1,power_law_1.2,3.2637439727783204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,4,balanced,0.3325600028038025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,4,balanced,0.3625226815541585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,4,balanced,0.37442131837209064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,4,balanced,0.373637318611145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,4,balanced,0.45977600415547687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,4,balanced,0.559551994005839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,2,1,power_law_1.2,9.907609558105468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,4,balanced,0.854751984278361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,2,power_law_1.2,34.66360168457031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,2,1,power_law_1.2,4.044249725341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,4,balanced,1.0659573078155518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,4,balanced,0.12287466724713643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,4,balanced,0.12591466307640076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,4,balanced,0.16642133394877115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,4,balanced,1.5368800163269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,2,1,power_law_1.01,0.7711359977722168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,4,balanced,0.16716800133387247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,4,balanced,0.16847467422485352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,4,balanced,0.17059733470280966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,4,balanced,0.17060265938440958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,4,balanced,0.1747679909070333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,4,balanced,1.9994293848673503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,4,balanced,0.18098666270573935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,4,balanced,8.14299201965332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,4,balanced,0.18405866622924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,4,balanced,0.19132800896962485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,4,balanced,0.20033599932988486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,2,1,power_law_1.2,4.859718322753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,4,balanced,0.2093013326327006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,4,balanced,0.23853333791097006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,4,balanced,2.9856160481770835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,4,balanced,0.2659466663996379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,2,1,power_law_1.01,21.137152099609374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,4,balanced,0.2974560062090556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,4,balanced,0.31411200761795044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,4,balanced,0.38489067554473877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,2,1,balanced,2.481621265411377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,4,balanced,0.4521546761194865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,4,balanced,0.6694133281707764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,4,balanced,4.074613253275554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,4,balanced,0.8292640050252279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,2,1,balanced,2.4950559933980307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,4,balanced,1.2219359874725342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,2,1,power_law_1.2,6.48431396484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,2,1,balanced,2.5156052907307944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,4,balanced,1.6180639266967773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,4,balanced,11.239456176757812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,2,1,balanced,2.550975958506266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,4,balanced,2.3802132606506348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,2,1,power_law_1.2,14.003071594238282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,4,balanced,7.076885223388672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,2,1,balanced,2.583418687184652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,4,balanced,3.10154660542806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,2,1,balanced,2.738821347554525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,2,1,power_law_1.2,8.107039642333984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,4,balanced,4.646170616149902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,4,balanced,10.13260269165039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,4,power_law_1.01,0.18465280532836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,4,power_law_1.01,0.22032639980316163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,4,power_law_1.01,0.32607359886169435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,4,balanced,6.1483306884765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,4,power_law_1.01,0.3483648061752319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,4,power_law_1.01,0.14485759735107423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,4,power_law_1.01,0.3499135971069336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,4,power_law_1.01,0.15694719552993774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,4,power_law_1.01,0.35068159103393554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,4,power_law_1.01,0.257203197479248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,4,power_law_1.01,0.3560703992843628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,4,power_law_1.01,0.2660608053207397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,4,power_law_1.01,0.3569216012954712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,4,power_law_1.01,0.2682111978530884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,4,power_law_1.01,0.36554880142211915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,4,power_law_1.01,0.2717056035995483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,4,power_law_1.01,0.3672703981399536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,4,power_law_1.01,0.2745280027389526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,4,power_law_1.01,0.37207679748535155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,4,power_law_1.01,0.38713600635528567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,2,1,power_law_1.01,27.663296508789063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,2,1,power_law_1.01,1.0782400131225587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,4,power_law_1.01,0.39686400890350343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,2,1,power_law_1.2,11.373036956787109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,4,power_law_1.01,0.41982078552246094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,4,power_law_1.01,0.12020479440689087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,4,power_law_1.01,0.1349120020866394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,4,power_law_1.01,0.4651199817657471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,2,1,power_law_1.01,1.7827135086059571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,4,power_law_1.01,0.15436160564422607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,2,1,balanced,2.707951863606771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,4,power_law_1.01,0.16505600214004518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,4,power_law_1.01,0.16592639684677124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,4,power_law_1.01,0.27866880893707274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,4,power_law_1.01,0.16745599508285522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,2,1,power_law_1.01,1.927987289428711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,4,power_law_1.01,0.5451519966125489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,4,power_law_1.01,0.2908351898193359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,4,power_law_1.01,0.17043839693069457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,4,power_law_1.01,0.29164159297943115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,4,power_law_1.01,0.6241024017333985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,4,power_law_1.01,0.3032128095626831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,2,1,power_law_1.01,2.035161590576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,4,power_law_1.01,0.7556352138519287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,2,1,power_law_1.2,18.245210266113283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,2,1,power_law_1.01,2.05743350982666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,4,power_law_1.01,0.17505279779434205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,4,power_law_1.01,1.0309120178222657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,4,power_law_1.01,0.18502399921417237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,2,1,power_law_1.01,2.0762304306030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,4,power_law_1.01,0.1906559944152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,4,power_law_1.01,0.3213248014450073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,4,power_law_1.01,0.20468480587005616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,4,power_law_1.01,0.3361471891403198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,2,1,power_law_1.2,14.633670043945312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,4,power_law_1.01,0.21992321014404298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,4,power_law_1.01,0.3672192096710205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,4,power_law_1.01,0.23692800998687744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,4,power_law_1.01,0.3982464075088501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,4,power_law_1.01,0.2671744108200073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,4,power_law_1.01,0.47481598854064944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,4,power_law_1.01,0.300166392326355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,4,power_law_1.01,0.5614848136901855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,4,power_law_1.01,0.32473599910736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,4,power_law_1.01,1.2712575912475585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,4,power_law_1.01,0.7245247840881348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,4,power_law_1.01,0.37706880569458007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,2,1,balanced,2.9132534662882485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,4,power_law_1.01,0.538489580154419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,4,power_law_1.01,0.7644991874694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,2,1,balanced,3.007509231567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,4,power_law_1.01,0.9727104187011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,4,balanced,0.04387733340263367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,4,balanced,0.04386133452256521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,4,balanced,0.06029333174228668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,4,balanced,0.0609493354956309
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,4,balanced,0.06343466540177663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,4,power_law_1.01,0.9942527770996094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,4,balanced,0.06504000226656596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,4,balanced,0.06603733201821645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,4,balanced,0.06890133519967397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,4,balanced,0.09920000036557515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,4,balanced,0.09663466612497966
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,4,balanced,0.09273599584897359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,4,balanced,0.22126932938893637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,4,balanced,0.21119999885559082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,4,balanced,0.17997332413991293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,4,balanced,0.1579039990901947
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,4,balanced,0.11115200320879619
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,4,balanced,0.08694932858149211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,4,balanced,0.10698666175206502
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,4,balanced,0.11275733510653178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,4,balanced,0.15243200461069742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,4,power_law_1.01,1.629030418395996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,4,power_law_1.01,1.3462207794189454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,4,power_law_1.01,1.3011455535888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,4,balanced,0.1787359913190206
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,4,balanced,0.24497065941492716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,4,power_law_1.01,1.993459129333496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,4,balanced,0.31148799260457355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,4,balanced,0.44704532623291016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,4,balanced,0.5782613356908163
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,4,balanced,0.8450720310211182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,4,power_law_1.01,2.616307258605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,4,balanced,1.1104319890340169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,4,power_law_1.01,1.7908287048339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,4,power_law_1.2,0.16174720525741576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,2,1,power_law_1.01,2.1951616287231444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,4,power_law_1.2,0.21057279109954835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,4,power_law_1.2,0.317305588722229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,4,power_law_1.2,0.34968318939208987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,4,power_law_1.01,2.7821184158325196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,4,power_law_1.2,0.3526463985443115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,2,1,power_law_1.01,2.2466047286987303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,4,power_law_1.2,0.3505215883255005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,4,power_law_1.2,0.35587201118469236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,4,power_law_1.2,0.3610431909561157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,4,power_law_1.2,0.36513280868530273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,2,1,power_law_1.2,21.212179565429686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,2,1,balanced,3.3561226526896157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,4,power_law_1.01,3.6239871978759766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,4,power_law_1.2,0.3694015979766846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,4,power_law_1.2,0.37263998985290525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,4,power_law_1.2,0.39893760681152346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,2,1,balanced,3.5316747029622397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,4,power_law_1.2,0.401094388961792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,4,power_law_1.01,2.3874048233032226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,4,power_law_1.2,0.44252800941467285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,2,1,power_law_1.01,2.313920021057129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,4,power_law_1.01,0.0446399986743927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,4,power_law_1.2,0.46300158500671384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,2,1,balanced,5.551839828491211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,2,1,power_law_1.01,2.412620735168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,4,power_law_1.2,0.4945216178894043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,2,1,balanced,5.834079742431641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,2,1,power_law_1.01,2.546278381347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,4,power_law_1.2,0.616319990158081
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,4,power_law_1.01,0.05549439787864685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,4,power_law_1.01,0.05058559775352478
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,4,power_law_1.01,0.057657599449157715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,4,power_law_1.01,0.06299520134925843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,2,1,power_law_1.01,2.69238395690918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,4,power_law_1.01,0.06369919776916504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,4,power_law_1.01,0.06503679752349853
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,4,power_law_1.01,0.06713600158691406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,4,power_law_1.2,0.9201408386230469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,4,power_law_1.01,0.08407040238380432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,4,power_law_1.01,3.4871105194091796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,2,1,balanced,8.505210876464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,4,power_law_1.01,0.08841599822044373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,4,power_law_1.01,0.08235520124435425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,4,power_law_1.01,0.15022079944610595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,4,power_law_1.01,4.20953598022461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,4,power_law_1.01,0.13133440017700196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,4,power_law_1.2,1.1300479888916015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,4,power_law_1.01,0.10300159454345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,4,power_law_1.01,0.12172160148620606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,2,1,balanced,9.245786666870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,4,power_law_1.01,0.14099199771881105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,4,power_law_1.01,7.112409973144532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,4,power_law_1.01,0.1374079942703247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,4,power_law_1.2,1.45032958984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,4,power_law_1.01,0.15230079889297485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,4,power_law_1.01,0.17239680290222167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,2,1,power_law_1.01,2.836787223815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,2,1,balanced,14.451119740804037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,4,power_law_1.01,0.2431936025619507
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,4,power_law_1.01,0.26841599941253663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,4,power_law_1.01,4.449548721313477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,4,power_law_1.2,1.80963191986084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,2,1,balanced,17.965866088867188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,4,power_law_1.2,0.1969151973724365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,4,power_law_1.2,0.20494720935821534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,4,power_law_1.2,0.27513599395751953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,4,power_law_1.2,0.26910080909729006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,2,1,power_law_1.01,2.9867904663085936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,4,power_law_1.2,0.27191040515899656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,4,power_law_1.2,0.2698048114776611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,4,power_law_1.2,0.27255680561065676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,4,power_law_1.2,0.28558080196380614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,2,1,power_law_1.2,27.748147583007814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,4,power_law_1.2,0.2867520093917847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,2,1,power_law_1.01,3.252742385864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,4,power_law_1.2,0.2990272045135498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,4,power_law_1.2,0.30159358978271483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,4,power_law_1.2,0.34155519008636476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,4,power_law_1.2,0.3531008005142212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,4,power_law_1.2,0.3845376014709473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,4,power_law_1.2,0.409119987487793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,4,power_law_1.01,9.109855651855469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,4,power_law_1.2,0.12055679559707641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,4,power_law_1.2,0.13587199449539183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,4,power_law_1.2,0.15497599840164183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,4,power_law_1.2,0.45258240699768065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,4,power_law_1.2,0.16512000560760498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,4,power_law_1.2,0.16606719493865968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,4,power_law_1.2,2.7062463760375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,4,power_law_1.2,0.1683135986328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,4,power_law_1.2,0.6260096073150635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,4,power_law_1.01,0.42490878105163576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,4,power_law_1.2,0.8366975784301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,4,power_law_1.2,0.17134079933166504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,4,power_law_1.2,0.1755328059196472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,4,power_law_1.2,0.18840960264205933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,4,power_law_1.2,0.1981760025024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,4,power_law_1.2,0.2083967924118042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,4,power_law_1.2,1.067859172821045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,4,power_law_1.01,5.142540740966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,4,power_law_1.2,0.23074560165405272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,4,power_law_1.2,0.23603200912475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,2,1,power_law_1.01,3.5665920257568358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,4,power_law_1.2,0.2629823923110962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,4,power_law_1.2,0.3146752119064331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,2,1,power_law_1.01,4.034419250488281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,4,power_law_1.2,0.3243200063705444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,4,power_law_1.2,0.4332608222961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,4,power_law_1.2,1.461350440979004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,2,1,power_law_1.01,5.007833480834961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,4,power_law_1.2,0.5617919921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,4,power_law_1.2,0.71113600730896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,2,1,power_law_1.01,6.0635841369628904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,4,power_law_1.01,0.4644927978515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,4,power_law_1.2,1.8982015609741212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,4,power_law_1.2,1.0275584220886231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,4,power_law_1.01,7.143820953369141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,2,1,power_law_1.01,7.977683258056641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,4,power_law_1.2,1.2784319877624513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,4,power_law_1.2,3.081644821166992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,2,1,power_law_1.01,9.939437103271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,4,power_law_1.2,1.819558334350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,4,power_law_1.2,4.271308898925781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,2,1,power_law_1.01,13.559770202636718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,4,power_law_1.01,0.7205440044403076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,4,power_law_1.2,3.4862590789794923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,4,power_law_1.01,6.851078033447266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,4,power_law_1.01,1.0101311683654786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,2,1,power_law_1.01,17.498323059082033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,4,power_law_1.2,2.6981760025024415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,4,power_law_1.01,9.407872009277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,4,1,balanced,0.07204266885916392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,4,1,balanced,0.0936853289604187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,4,1,balanced,0.15549866358439127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,4,1,balanced,0.161189337571462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,4,1,balanced,0.168122669061025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,4,1,balanced,0.16841065883636475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,4,1,balanced,0.16941332817077637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,4,1,balanced,0.1710933248202006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,4,1,balanced,0.17284266153971353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,4,1,balanced,0.17203199863433838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,4,1,balanced,0.17646400133768717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,4,1,balanced,0.18133866786956787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,4,1,balanced,0.18513067563374838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,4,1,balanced,0.19197332859039307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,4,1,balanced,0.25117866198221844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,4,1,balanced,0.25645333528518677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,4,1,balanced,0.2622026602427165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,4,1,balanced,0.37437868118286133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,4,1,balanced,0.45392000675201416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,4,1,balanced,0.6443093220392863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,4,1,balanced,0.8139359951019287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,4,1,balanced,1.1868639787038167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,4,1,balanced,1.5464426676432292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,4,power_law_1.2,6.716102600097656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,4,1,balanced,2.3337119420369468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,4,1,balanced,3.133376121520996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,4,1,balanced,4.942106564839681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,4,1,balanced,6.652010599772136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,4,1,balanced,0.07045866549015045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,4,1,balanced,0.08477333188056946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,4,power_law_1.2,0.04663040041923523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,4,1,balanced,0.1267466644446055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,4,1,balanced,0.1334826648235321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,4,1,balanced,0.1385599970817566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,4,1,balanced,0.1397599975268046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,4,1,balanced,0.14448533455530801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,4,1,balanced,0.14030399918556213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,4,1,balanced,0.15101866920789084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,4,1,balanced,0.15039466818173727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,4,1,balanced,0.16049066185951233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,4,1,balanced,0.1660373310248057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,4,1,balanced,0.1732106606165568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,4,power_law_1.2,5.3065025329589846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,4,1,balanced,0.1820746660232544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,4,1,balanced,0.2488480011622111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,4,1,balanced,0.2656373381614685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,4,1,balanced,0.264080007870992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,4,1,balanced,0.3619519869486491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,4,1,balanced,0.44706134001413983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,4,1,balanced,0.6356906493504842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,4,power_law_1.01,13.074041748046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,4,power_law_1.2,0.05063679814338684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,4,1,balanced,0.817408005396525
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,4,power_law_1.2,0.05286399722099304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,4,power_law_1.2,0.05829120278358459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,4,1,balanced,1.1799840132395427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,4,power_law_1.2,0.0642687976360321
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,4,power_law_1.2,0.06637439727783204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,4,1,balanced,1.5502773920694988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,4,power_law_1.2,0.06670719981193543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,4,power_law_1.2,0.06936960220336914
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,4,power_law_1.2,0.07203199863433837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,4,1,balanced,2.2988533973693848
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,4,power_law_1.2,0.07970560193061829
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,4,power_law_1.2,0.07649919986724854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,4,power_law_1.2,0.1506495952606201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,4,1,balanced,3.0348854064941406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,4,power_law_1.2,0.12357759475708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,4,power_law_1.2,0.11030399799346924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,4,1,balanced,4.5254561106363935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,4,power_law_1.2,0.15332479476928712
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,4,power_law_1.2,0.12361600399017333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,4,1,balanced,6.009717305501302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,4,1,balanced,0.058506667613983154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,4,1,balanced,0.07242666681607564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,4,1,balanced,0.10699199636777242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,4,1,balanced,0.10564800103505452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,4,1,balanced,0.1062453289826711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,4,1,balanced,0.1072746713956197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,4,1,balanced,0.1072266697883606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,4,1,balanced,0.11014399925867717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,4,1,balanced,0.11121599872907002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,4,1,balanced,0.11342400312423706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,4,1,balanced,0.11659733454386394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,4,1,balanced,0.12172266840934753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,4,1,balanced,0.12664000193277994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,4,1,balanced,0.13871999581654867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,4,1,balanced,0.1923253337542216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,4,1,balanced,0.1967573364575704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,4,1,balanced,0.20214933156967163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,4,1,balanced,0.2656853397687276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,4,1,balanced,0.3350079854329427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,4,1,balanced,0.4749386707941691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,4,1,balanced,0.5953173240025839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,4,1,balanced,0.8726560274759928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,4,1,balanced,1.1309066613515217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,4,1,balanced,1.6687253316243489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,4,1,balanced,2.181429386138916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,4,power_law_1.2,0.14621440172195435
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,4,power_law_1.2,0.14465919733047486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,4,1,balanced,3.2692267100016275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,4,1,balanced,4.308986663818359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,4,power_law_1.2,0.16877440214157105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,4,power_law_1.2,4.011520004272461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,4,power_law_1.01,1.4256383895874023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,4,power_law_1.01,17.793612670898437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,4,power_law_1.2,9.489177703857422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,4,1,balanced,0.027237333357334137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,4,1,balanced,0.035162667433420815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,4,1,balanced,0.057258665561676025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,4,1,balanced,0.057855998476346336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,4,1,balanced,0.05762133498986562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,4,1,balanced,0.06010666489601135
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,4,1,balanced,0.0645066648721695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,4,1,balanced,0.06611200173695882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,4,1,balanced,0.06703466673692067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,4,1,balanced,0.06779199838638306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,4,1,balanced,0.06905599931875865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,4,1,balanced,0.07096000015735626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,4,1,balanced,0.07111466427644093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,4,1,balanced,0.07460266848405202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,4,1,balanced,0.11850666999816895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,4,1,balanced,0.10085333387056987
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,4,1,balanced,0.08618666728337605
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,4,1,balanced,0.1883359948794047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,4,1,balanced,0.11111467083295186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,4,1,balanced,0.14005866646766663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,4,1,balanced,0.1690559983253479
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,4,1,balanced,0.23220266898473105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,4,1,balanced,0.2940000096956889
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,4,1,balanced,0.4203733205795288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,4,1,balanced,0.5525439977645874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,4,1,balanced,0.8066399892171224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,4,power_law_1.2,0.2230976104736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,4,1,balanced,1.057919979095459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,4,1,power_law_1.01,0.0716480016708374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,4,1,power_law_1.01,0.08658559918403626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,4,1,power_law_1.01,0.10012160539627075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,4,1,power_law_1.01,0.1440832018852234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,4,1,power_law_1.01,0.1633407950401306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,4,1,power_law_1.01,0.16741119623184203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,4,1,power_law_1.01,0.16866559982299806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,4,1,power_law_1.01,0.16957440376281738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,4,1,power_law_1.01,0.18750079870223998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,4,1,power_law_1.01,0.19720319509506226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,4,power_law_1.2,0.2969536066055298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,4,power_law_1.2,8.520671844482422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,4,1,power_law_1.01,0.21960959434509278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,4,1,power_law_1.01,0.22570879459381105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,4,1,power_law_1.01,0.2294015884399414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,4,1,power_law_1.01,0.24588160514831542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,4,1,power_law_1.01,0.256876802444458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,4,1,power_law_1.01,0.2976831912994385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,4,1,power_law_1.01,0.33803520202636717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,4,1,power_law_1.01,0.4178368091583252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,4,power_law_1.2,5.111801528930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,4,1,power_law_1.01,0.5049856185913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,4,power_law_1.2,0.3901632070541382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,4,1,power_law_1.01,0.7051968097686767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,4,power_law_1.01,23.798361206054686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,4,1,power_law_1.01,0.06327679753303528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,4,1,power_law_1.01,0.07103999853134155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,4,1,power_law_1.01,0.08225280046463013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,4,1,power_law_1.01,0.11169279813766479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,4,1,power_law_1.01,0.13141759634017944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,4,1,power_law_1.01,0.13116159439086914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,4,1,power_law_1.01,0.13965439796447754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,4,1,power_law_1.01,0.8711296081542969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,4,1,power_law_1.01,0.14403200149536133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,4,1,power_law_1.01,0.15674879550933837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,4,1,power_law_1.01,0.16525440216064452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,4,1,power_law_1.01,0.17269760370254517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,4,1,power_law_1.01,0.1932031989097595
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,4,power_law_1.01,1.875155258178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,4,power_law_1.2,0.5798336029052734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,4,1,power_law_1.01,0.1956544041633606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,4,1,power_law_1.01,0.060838401317596436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,4,1,power_law_1.01,0.22726399898529054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,4,1,power_law_1.01,0.06873599886894226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,4,1,power_law_1.01,0.23907840251922607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,4,1,power_law_1.01,0.07620480060577392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,4,1,power_law_1.01,0.10494719743728638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,4,1,power_law_1.01,0.10917119979858399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,4,power_law_1.01,17.599122619628908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,4,1,power_law_1.01,0.10613119602203369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,4,1,power_law_1.01,0.10678399801254272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,4,1,power_law_1.01,1.2445247650146485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,4,1,power_law_1.01,0.10845439434051514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,4,1,power_law_1.01,0.11034239530563354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,4,1,power_law_1.01,0.11244159936904907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,4,1,power_law_1.01,0.12050559520721435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,4,1,power_law_1.01,0.13953280448913574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,4,1,power_law_1.01,0.15302400588989257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,4,1,power_law_1.01,1.6239936828613282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,4,1,power_law_1.01,0.16764800548553466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,4,1,power_law_1.01,0.17766400575637817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,4,1,power_law_1.01,0.19785599708557128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,4,1,power_law_1.01,0.27500159740448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,4,1,power_law_1.01,0.2191231966018677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,4,1,power_law_1.01,0.3031232118606567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,4,1,power_law_1.01,0.28523519039154055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,4,1,power_law_1.01,0.4010047912597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,4,power_law_1.2,0.7151423931121826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,4,1,power_law_1.01,0.49257597923278806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,4,1,power_law_1.01,0.028998398780822755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,4,1,power_law_1.01,0.03596160113811493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,4,1,power_law_1.01,0.35469439029693606
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,4,1,power_law_1.01,0.045721599459648134
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,4,1,power_law_1.01,0.0559935986995697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,4,1,power_law_1.01,0.05759360194206238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,4,1,power_law_1.01,0.05924479961395264
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,4,1,power_law_1.01,0.06354560256004334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,4,1,power_law_1.01,0.0690559983253479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,4,1,power_law_1.01,0.48122878074645997
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,4,1,power_law_1.01,0.06395519971847534
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,4,1,power_law_1.01,0.06829439997673034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,4,1,power_law_1.01,0.0726527988910675
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,4,1,power_law_1.01,0.0699072003364563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,4,1,power_law_1.01,0.07583360075950622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,4,1,power_law_1.01,0.6243840217590332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,4,1,power_law_1.01,0.6734272003173828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,4,1,power_law_1.01,0.0823743999004364
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,4,1,power_law_1.01,0.08213760256767273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,4,1,power_law_1.01,0.08704000115394592
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,4,1,power_law_1.01,0.09559040069580078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,4,1,power_law_1.01,0.1083840012550354
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,4,1,power_law_1.01,0.12445440292358398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,4,1,power_law_1.01,0.8912447929382324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,4,1,power_law_1.01,2.363212776184082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,4,1,power_law_1.01,0.8511103630065918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,4,1,power_law_1.01,0.16338560581207276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,4,1,power_law_1.01,1.1573887825012208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,4,1,power_law_1.01,1.2362815856933593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,4,1,power_law_1.01,0.18738559484481812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,4,1,power_law_1.01,0.24154880046844482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,4,1,power_law_1.01,3.109324836730957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,4,1,power_law_1.01,1.5917311668395997
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,4,1,power_law_1.01,0.30513920783996584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,4,1,power_law_1.01,0.43916797637939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,4,1,power_law_1.01,2.328825569152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,4,power_law_1.2,1.1242112159729003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,4,1,power_law_1.01,4.708966445922852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,4,1,power_law_1.01,1.6890432357788085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,4,power_law_1.2,7.1791740417480465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,4,1,power_law_1.01,0.562937593460083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,4,1,power_law_1.01,3.0784832000732423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,4,1,power_law_1.01,2.2312192916870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,4,power_law_1.2,12.70914535522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,4,power_law_1.2,16.802029418945313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,4,1,power_law_1.01,6.759276580810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,4,1,power_law_1.2,0.0773248016834259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,4,1,power_law_1.2,0.08768640160560608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,4,1,power_law_1.2,0.10153599977493286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,4,1,power_law_1.2,0.1419584035873413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,4,1,power_law_1.2,0.16700799465179444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,4,1,power_law_1.2,0.16944639682769774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,4,1,power_law_1.2,0.16904319524765016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,4,1,power_law_1.2,0.1677248001098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,4,1,power_law_1.2,0.179750394821167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,4,1,power_law_1.2,0.2066431999206543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,4,1,power_law_1.2,0.21303040981292726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,4,1,power_law_1.2,0.22430078983306884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,4,1,power_law_1.2,0.22885758876800538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,4,1,power_law_1.2,0.24465279579162597
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,4,1,power_law_1.01,0.8196991920471192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,4,1,power_law_1.2,0.25655040740966795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,4,1,power_law_1.2,0.30580480098724366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,4,1,power_law_1.2,0.3416064023971558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,4,1,power_law_1.2,0.4313727855682373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,4,1,power_law_1.2,0.5036159992218018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,4,1,power_law_1.01,3.282291030883789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,4,1,power_law_1.2,0.6999743938446045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,4,1,power_law_1.2,0.875391960144043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,4,1,power_law_1.01,4.554707336425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,4,power_law_1.2,20.66768035888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,4,1,power_law_1.2,0.06268799901008607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,4,1,power_law_1.2,0.07110400199890136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,4,1,power_law_1.2,0.08729599714279175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,4,1,power_law_1.2,0.11458560228347778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,4,1,power_law_1.2,0.12840319871902467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,4,1,power_law_1.2,0.13344639539718628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,4,1,power_law_1.2,0.13873280286788942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,4,1,power_law_1.2,0.1432703971862793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,4,1,power_law_1.2,0.17214720249176024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,4,1,power_law_1.2,0.16242560148239135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,4,1,power_law_1.2,0.1817023992538452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,4,1,power_law_1.2,0.19199999570846557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,4,1,power_law_1.2,0.20091519355773926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,4,1,power_law_1.01,4.353241729736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,4,1,power_law_1.2,0.060012799501419065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,4,1,power_law_1.2,0.22119040489196778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,4,1,power_law_1.2,0.06899200081825256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,4,1,power_law_1.2,0.08689919710159302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,4,1,power_law_1.2,0.10417920351028442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,4,1,power_law_1.2,0.24187519550323486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,4,1,power_law_1.2,0.10412800312042236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,4,1,power_law_1.2,0.10597759485244751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,4,1,power_law_1.2,0.10778239965438843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,4,1,power_law_1.2,0.10987520217895508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,4,1,power_law_1.2,0.11034239530563354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,4,1,power_law_1.2,0.11148159503936768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,4,1,power_law_1.2,0.2691263914108276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,4,1,power_law_1.2,0.11829119920730591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,4,1,power_law_1.2,0.13337600231170654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,4,1,power_law_1.2,1.2405376434326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,4,1,power_law_1.2,0.1510591983795166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,4,1,power_law_1.2,0.16664960384368896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,4,1,power_law_1.2,0.30398719310760497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,4,1,power_law_1.2,0.17927039861679078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,4,1,power_law_1.2,0.19767040014266968
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,4,1,power_law_1.01,1.095404815673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,4,1,power_law_1.2,0.40164480209350584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,4,1,power_law_1.2,0.21923840045928955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,4,1,power_law_1.2,0.48779520988464353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,4,1,power_law_1.2,0.2894848108291626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,4,power_law_1.2,1.7564224243164062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,4,1,power_law_1.2,0.3475712060928345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,4,1,power_law_1.2,0.03054719865322113
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,4,1,power_law_1.2,0.03495680093765259
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,4,1,power_law_1.2,0.042361599206924436
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,4,1,power_law_1.2,0.05619199872016907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,4,1,power_law_1.2,0.05628799796104431
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,4,1,power_law_1.2,0.6730944156646729
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,4,1,power_law_1.2,0.05934079885482788
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,4,1,power_law_1.2,0.06378239989280701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,4,1,power_law_1.2,0.06984320282936096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,4,1,power_law_1.2,0.06511359810829162
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,4,1,power_law_1.2,0.06807039976119995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,4,1,power_law_1.2,0.4861311912536621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,4,1,power_law_1.2,0.07193599939346314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,4,1,power_law_1.2,0.07143679857254029
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,4,1,power_law_1.2,0.07464320063591004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,4,1,power_law_1.2,0.08173440098762512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,4,1,power_law_1.2,0.0797439992427826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,4,1,power_law_1.2,0.08407679796218873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,4,1,power_law_1.2,0.09338240027427673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,4,1,power_law_1.2,1.605753517150879
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,4,1,power_law_1.2,0.11542400121688842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,4,1,power_law_1.2,0.8543935775756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,4,power_law_1.2,19.860890197753907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,4,balanced,0.2166986664136251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,4,balanced,0.21719467639923096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,4,balanced,0.4785706599553426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,4,balanced,0.4750986496607463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,4,balanced,0.4773333470026652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,4,balanced,0.4773919979731242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,4,balanced,0.4769493341445923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,4,1,power_law_1.2,0.12574080228805543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,4,1,power_law_1.2,0.6298111915588379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,4,1,power_law_1.2,1.2356032371520995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,4,balanced,0.48054933547973633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,4,balanced,0.48604798316955566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,4,balanced,0.49089598655700684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,4,balanced,0.4970080057779948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,4,balanced,0.5072746674219767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,4,balanced,0.5085493326187134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,4,balanced,0.519541343053182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,4,1,power_law_1.2,0.16310399770736694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,4,balanced,0.5737866560618082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,4,balanced,0.5812693436940511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,4,balanced,0.5867786804835001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,4,balanced,0.6986613273620605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,4,balanced,0.8186826705932617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,4,balanced,1.3397173881530762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,4,balanced,1.6363946596781414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,4,balanced,2.512495994567871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,4,balanced,3.4977760314941406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,4,1,power_law_1.2,0.187283194065094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,4,balanced,5.718394597371419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,4,1,power_law_1.2,1.5974847793579101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,4,1,power_law_1.2,0.8804032325744628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,4,1,power_law_1.2,0.2547904014587402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,4,balanced,8.012592315673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,4,1,power_law_1.01,6.044851303100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,4,balanced,0.18668266137441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,4,balanced,0.18689600626627603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,4,balanced,0.3482186794281006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,4,balanced,0.35263999303181964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,4,balanced,0.34988800684611004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,4,balanced,0.3539573351542155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,4,1,power_law_1.2,2.347430419921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,4,balanced,0.3569759925206502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,4,1,power_law_1.2,2.354252815246582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,4,balanced,0.35672001043955487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,4,balanced,0.36187732219696045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,4,balanced,0.37034134070078534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,4,balanced,0.37196266651153564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,4,balanced,12.8940798441569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,4,balanced,0.3829600016276042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,4,balanced,0.3970079819361369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,4,balanced,0.41682132085164386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,4,balanced,0.4758613506952922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,4,1,power_law_1.2,0.30423679351806643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,4,power_law_1.2,9.821292877197266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,4,balanced,0.15041066209475198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,4,balanced,0.1639840006828308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,4,balanced,0.20956265926361084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,4,balanced,0.21211733420689902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,4,balanced,0.21226133902867636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,4,balanced,0.2153279980023702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,4,balanced,0.21838400761286417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,4,balanced,0.22040534019470215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,4,balanced,0.2295680046081543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,4,balanced,0.23312000433603922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,4,balanced,0.23894399404525757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,4,balanced,0.4967840115229289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,4,balanced,0.5263359944025675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,4,balanced,0.6604959964752197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,4,balanced,0.846186637878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,4,balanced,1.2394293149312336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,4,balanced,1.5819306373596191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,4,balanced,18.061242421468098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,4,balanced,0.2558186650276184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,4,balanced,0.2601066629091899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,4,balanced,2.330597400665283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,4,balanced,0.2914186716079712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,4,balanced,0.34810133775075275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,4,balanced,0.3657279809315999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,4,balanced,0.38577600320180255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,4,balanced,0.5060266653696696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,4,balanced,3.064922650655111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,4,balanced,0.6616319815317789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,4,balanced,0.8878933588663737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,4,balanced,1.1739786465962727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,4,balanced,4.629941304524739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,4,balanced,1.698080062866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,4,balanced,2.277466615041097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,4,balanced,6.5272642771403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,4,balanced,3.3449227015177407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,4,power_law_1.2,1.994246482849121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,4,1,power_law_1.2,1.1499967575073242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,4,balanced,11.267876942952475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,4,power_law_1.01,0.21407361030578614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,4,balanced,4.41651725769043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,4,power_law_1.01,0.27898879051208497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,4,power_law_1.01,0.43179521560668943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,4,power_law_1.01,0.4738880157470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,4,power_law_1.01,0.4766079902648926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,4,power_law_1.01,0.4797311782836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,4,power_law_1.01,0.48020482063293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,4,balanced,6.612885157267253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,4,power_law_1.01,0.4831808090209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,4,power_law_1.01,0.49061760902404783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,4,power_law_1.01,0.49158401489257814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,4,power_law_1.01,0.49660158157348633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,4,power_law_1.01,0.5495039939880371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,4,1,power_law_1.2,3.095033645629883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,4,power_law_1.01,0.5606143951416016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,4,balanced,16.37833023071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,4,power_law_1.01,0.6134848117828369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,4,balanced,8.76904551188151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,4,power_law_1.01,0.19402240514755248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,4,power_law_1.01,0.6560063838958741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,4,power_law_1.01,0.2126847982406616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,4,power_law_1.01,0.15619839429855348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,4,power_law_1.01,0.3261823892593384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,4,power_law_1.01,0.20357120037078857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,4,power_law_1.01,0.34620161056518556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,4,power_law_1.01,0.20830719470977782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,4,power_law_1.01,0.3464256048202515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,4,power_law_1.01,0.20999679565429688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,4,power_law_1.01,0.3518208026885986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,4,power_law_1.01,0.21082239151000975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,4,power_law_1.01,0.21349759101867677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,4,power_law_1.01,0.21777920722961425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,4,power_law_1.01,0.36036479473114014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,4,power_law_1.01,0.2220736026763916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,4,power_law_1.01,0.3765696048736572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,4,power_law_1.01,0.24378879070281984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,4,power_law_1.01,0.3796864032745361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,4,power_law_1.01,0.2520639896392822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,4,power_law_1.01,0.3924288034439087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,4,power_law_1.01,0.2708159923553467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,4,power_law_1.01,0.7537792205810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,4,power_law_1.01,0.2941567897796631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,4,power_law_1.01,0.4103871822357178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,4,power_law_1.01,0.3094144105911255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,4,power_law_1.01,0.47730560302734376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,4,power_law_1.01,0.3454528093338013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,4,power_law_1.01,1.0053248405456543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,4,power_law_1.01,0.4929215908050537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,4,power_law_1.01,0.5193280220031739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,4,power_law_1.01,0.41904640197753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,4,power_law_1.01,1.2486528396606444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,4,power_law_1.01,0.6028096199035644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,4,power_law_1.01,0.4961728096008301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,4,power_law_1.01,0.6577280044555665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,4,1,power_law_1.2,3.0848255157470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,4,power_law_1.01,0.5974592208862305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,4,power_law_1.01,1.59236478805542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,4,power_law_1.01,0.81844482421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,4,1,power_law_1.2,0.4421055793762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,4,power_law_1.01,0.8563263893127442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,4,power_law_1.01,1.0910400390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,4,balanced,0.058592001597086586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,4,balanced,0.05885333319505056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,4,balanced,0.07973333199818929
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,4,balanced,0.08032000064849854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,4,balanced,0.08283199866612752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,4,balanced,0.0835093359152476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,4,balanced,0.087226668993632
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,4,balanced,0.08927999933560689
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,4,balanced,0.12340799967447917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,4,balanced,0.11961066722869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,4,balanced,0.11215466260910034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,4,power_law_1.01,1.5667136192321778
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,4,balanced,0.2892586588859558
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,4,balanced,0.2762879927953084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,4,balanced,0.23798400163650513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,4,balanced,0.19954667488733926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,4,balanced,0.1323253313700358
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,4,balanced,0.11033067107200623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,4,balanced,0.16332800189654031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,4,balanced,0.16806934277216592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,4,power_law_1.01,1.0253824234008788
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,4,balanced,0.2107093334197998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,4,balanced,0.2718293269475301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,4,power_law_1.01,2.6527168273925783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,4,balanced,0.3793226480484009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,4,power_law_1.01,2.4275455474853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,4,balanced,0.4994560082753499
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,4,balanced,0.7366453011830648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,4,balanced,0.9740853309631348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,4,balanced,1.4563892682393391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,4,balanced,1.947114626566569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,4,power_law_1.01,1.448089599609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,4,power_law_1.01,2.6205440521240235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,4,power_law_1.01,2.919424057006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,4,power_law_1.01,1.869011116027832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,4,power_law_1.01,4.152179336547851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,4,power_law_1.01,4.1226753234863285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,4,1,power_law_1.2,1.6889856338500977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,4,power_law_1.01,5.649939346313476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,4,power_law_1.01,2.9504512786865233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,4,1,power_law_1.2,0.5675136089324951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,4,1,power_law_1.2,4.565318298339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,4,1,power_law_1.2,4.580108642578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,4,power_law_1.01,5.645465469360351
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,4,power_law_1.01,0.0548799991607666
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,4,power_law_1.01,0.07323520183563233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,4,power_law_1.01,0.06839039921760559
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,4,power_law_1.01,0.07738239765167236
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,4,power_law_1.01,0.08711680173873901
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,4,power_law_1.01,0.08595839738845826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,4,power_law_1.01,0.08599680066108703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,4,power_law_1.01,0.0905344009399414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,4,power_law_1.01,0.10508160591125489
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,4,power_law_1.01,0.11098239421844483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,4,power_law_1.01,0.10487040281295776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,4,power_law_1.01,0.14815360307693481
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,4,power_law_1.01,0.16299519538879395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,4,power_law_1.01,0.1506943941116333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,4,1,power_law_1.2,2.2100479125976564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,4,power_law_1.01,0.1841599941253662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,4,power_law_1.01,0.16744320392608641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,4,power_law_1.01,3.614451217651367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,4,power_law_1.01,0.1745471954345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,4,power_law_1.01,0.21709439754486085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,4,power_law_1.01,0.28618879318237306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,4,power_law_1.01,0.3439296007156372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,4,power_law_1.01,8.980550384521484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,4,power_law_1.01,0.5001088142395019
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,4,power_law_1.01,0.6841343879699707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,4,power_law_1.01,11.374578857421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,4,power_law_1.01,5.244473648071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,4,power_law_1.01,12.416627502441406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,4,power_law_1.01,0.8853952407836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,4,power_law_1.01,14.913075256347657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,4,power_law_1.01,1.4066944122314453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,4,power_law_1.01,20.353152465820312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,4,power_law_1.01,6.858150482177734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,4,1,power_law_1.2,6.673299407958984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,4,power_law_1.2,0.24686079025268554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,4,power_law_1.2,0.302291202545166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,4,power_law_1.2,0.4217535972595215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,4,power_law_1.2,0.4786111831665039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,4,power_law_1.2,0.4730559825897217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,4,power_law_1.2,0.47619199752807617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,4,power_law_1.01,1.9081279754638671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,4,power_law_1.2,0.47830400466918943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,4,power_law_1.2,0.481766414642334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,4,power_law_1.2,0.48635520935058596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,4,power_law_1.2,0.492633581161499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,4,power_law_1.2,0.5050559997558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,4,power_law_1.2,0.5462528228759765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,4,1,power_law_1.2,3.2866752624511717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,4,power_law_1.2,0.5433472156524658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,4,power_law_1.2,0.6187839984893799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,4,power_law_1.2,0.6421504020690918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,4,power_law_1.2,0.7312831878662109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,4,power_law_1.2,1.0024959564208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,4,power_law_1.2,1.306208038330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,4,power_law_1.01,28.448504638671874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,4,1,power_law_1.2,0.8217023849487305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,4,power_law_1.2,1.8830720901489257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,4,power_law_1.2,2.3966079711914063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,4,power_law_1.01,11.18252182006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,4,power_law_1.01,2.8205503463745116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,4,power_law_1.2,3.063212776184082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,4,1,power_law_1.2,6.071775817871094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,4,power_law_1.2,0.18591359853744507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,4,power_law_1.2,0.22196478843688966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,4,power_law_1.2,0.3223680019378662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,4,power_law_1.2,0.3448319911956787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,4,power_law_1.2,0.34478719234466554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,4,power_law_1.2,0.349017596244812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,4,power_law_1.2,0.35715839862823484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,4,power_law_1.2,0.3710592031478882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,4,power_law_1.2,0.3773439884185791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,4,power_law_1.2,4.893286514282226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,4,power_law_1.2,0.3923392057418823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,4,power_law_1.2,0.42062082290649416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,4,power_law_1.2,0.44893441200256345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,4,power_law_1.2,0.5005951881408691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,4,power_law_1.2,0.5102015972137451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,4,power_law_1.2,0.6135871887207032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,4,power_law_1.2,0.7614399909973144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,4,power_law_1.2,0.8190336227416992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,4,power_law_1.2,1.1711487770080566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,4,power_law_1.2,1.4256447792053222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,4,power_law_1.2,6.367622375488281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,4,power_law_1.01,31.092019653320314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,4,power_law_1.2,0.1567296028137207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,4,power_law_1.2,0.187827205657959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,4,power_law_1.2,0.1940160036087036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,4,power_law_1.2,2.133459281921387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,4,power_law_1.2,0.21066880226135254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,4,power_law_1.2,0.21257600784301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,4,power_law_1.2,0.2145792007446289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,4,power_law_1.2,0.21930880546569825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,4,power_law_1.2,0.22408320903778076
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,4,power_law_1.01,3.8903873443603514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,4,power_law_1.2,2.925030326843262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,4,power_law_1.2,0.24450559616088868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,4,power_law_1.2,0.2473599910736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,4,power_law_1.2,0.2699264049530029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,4,power_law_1.2,0.2813055992126465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,4,power_law_1.2,0.3177727937698364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,4,power_law_1.2,0.3682624101638794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,4,power_law_1.2,0.4287744045257568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,4,power_law_1.2,0.4773439884185791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,4,power_law_1.2,0.6065792083740235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,4,power_law_1.2,0.8547200202941895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,4,power_law_1.2,1.0385472297668457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,4,power_law_1.2,5.411705780029297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,4,power_law_1.2,1.5327232360839844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,4,power_law_1.01,37.851007080078126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,4,1,balanced,0.09913067022959392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,4,1,balanced,0.13662933309872946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,4,1,balanced,0.25486934185028076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,4,power_law_1.2,1.9670719146728515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,4,power_law_1.2,9.329747009277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,4,power_law_1.2,0.05479679703712463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,4,power_law_1.2,0.0642624020576477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,4,power_law_1.2,0.07248640060424805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,4,power_law_1.2,0.07730559706687927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,4,1,balanced,0.25754666328430176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,4,power_law_1.2,0.08232960104942322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,4,1,balanced,0.2564586599667867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,4,1,balanced,0.25883734226226807
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,4,power_law_1.2,0.0902783989906311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,4,1,balanced,0.262447992960612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,4,power_law_1.2,0.09320319890975952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,4,power_law_1.2,0.09843840003013611
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,4,power_law_1.2,0.10394879579544067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,4,power_law_1.2,0.10728319883346557
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,4,power_law_1.2,0.10720640420913696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,4,power_law_1.2,0.16244479417800903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,4,power_law_1.2,0.14532480239868165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,4,1,balanced,0.2624640067418416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,4,1,balanced,0.2648426691691081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,4,power_law_1.2,6.153830337524414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,4,1,balanced,0.27245867252349854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,4,power_law_1.2,0.11845120191574096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,4,1,balanced,0.27391467491785687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,4,1,balanced,0.2760480046272278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,4,1,balanced,0.2772266666094462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,4,1,balanced,0.31223465998967487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,4,1,balanced,0.42181865374247235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,4,power_law_1.2,0.15564800500869752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,4,1,balanced,0.42505598068237305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,4,power_law_1.2,0.17219200134277343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,4,power_law_1.2,0.16403839588165284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,4,1,balanced,0.4249546527862549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,4,1,balanced,0.5997493267059326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,4,1,balanced,0.7172373135884603
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,4,power_law_1.2,0.2392575979232788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,4,1,balanced,1.0615999698638916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,4,1,balanced,1.324127991994222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,4,power_law_1.2,2.902943992614746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,4,1,balanced,1.978826681772868
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,4,power_law_1.2,0.2743040084838867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,4,power_law_1.2,0.3572671890258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,4,1,balanced,2.734560012817383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,4,1,balanced,4.231568018595378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,4,1,power_law_1.2,4.347596740722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,4,1,balanced,0.08515200018882751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,4,1,balanced,0.10314666231473286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,4,1,balanced,0.18230400482813516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,4,1,balanced,0.1839039921760559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,4,1,balanced,5.566677093505859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,4,1,balanced,0.1711039940516154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,4,1,balanced,0.17319466670354208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,4,1,balanced,0.17683200041453043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,4,1,balanced,0.1788853406906128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,4,1,balanced,0.18888000647226968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,4,1,balanced,0.18971733252207437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,4,1,balanced,0.19852266709009805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,4,1,balanced,0.2103360096613566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,4,1,balanced,0.2223680019378662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,4,1,balanced,0.2659839987754822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,4,1,balanced,0.3595786492029826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,4,1,balanced,0.3737066586812337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,4,1,balanced,0.3948479890823364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,4,1,balanced,0.548469344774882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,4,1,balanced,0.6910506884256998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,4,1,balanced,8.522672017415365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,4,1,balanced,1.0077706972757976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,4,1,balanced,1.3023093541463215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,4,power_law_1.2,13.944038391113281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,4,1,balanced,1.9082026481628418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,4,power_law_1.2,0.47845120429992677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,4,1,balanced,2.510741392771403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,4,1,balanced,3.7310508092244468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,4,1,balanced,13.546618143717447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,4,1,balanced,0.08029333253701527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,4,1,balanced,0.09948800007502238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,4,1,balanced,0.16821332772572836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,4,1,balanced,4.949621200561523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,4,1,balanced,0.16920000314712524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,4,1,balanced,0.16877333323160806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,4,1,balanced,0.16819733381271362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,4,1,balanced,0.1706613302230835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,4,1,balanced,0.17172799507776895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,4,1,balanced,0.17492266496022543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,4,1,balanced,0.17595734198888144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,4,power_law_1.2,4.281536102294922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,4,1,balanced,0.17876267433166504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,4,power_law_1.2,8.771263885498048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,4,1,balanced,0.18522133429845175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,4,1,balanced,0.19047466913859049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,4,1,balanced,0.21090133984883627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,4,1,balanced,0.27400533358256024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,4,1,balanced,7.3854719797770185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,4,1,balanced,0.2844533324241638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,4,1,balanced,0.2914239962895711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,4,1,balanced,0.3937813440958659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,4,1,balanced,0.48589332898457843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,4,1,balanced,0.7030239899953207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,4,1,balanced,0.8926719824473063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,4,1,balanced,1.3163093725840251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,4,1,balanced,1.7180693944295247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,4,1,balanced,9.814704259236654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,4,1,balanced,2.5503199895222983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,4,1,balanced,3.3415679931640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,4,1,balanced,5.002634684244792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,4,power_law_1.2,0.8010496139526367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,4,1,power_law_1.2,1.0963775634765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,4,1,power_law_1.01,0.0963648021221161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,4,1,power_law_1.01,0.117740797996521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,4,1,balanced,6.62502924601237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,4,1,power_law_1.01,0.14135040044784547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,4,1,power_law_1.01,0.08526080250740051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,4,1,power_law_1.01,0.25155200958251955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,4,1,power_law_1.01,0.1003648042678833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,4,1,power_law_1.01,0.25080959796905516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,4,1,power_law_1.01,0.12037760019302368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,4,1,power_law_1.01,0.248473596572876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,4,1,power_law_1.01,0.1735103964805603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,4,1,power_law_1.01,0.2528512001037598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,4,1,power_law_1.01,0.2566783905029297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,4,1,power_law_1.01,0.27946879863739016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,4,1,power_law_1.01,0.31281280517578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,4,1,power_law_1.01,0.33189759254455564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,4,power_law_1.2,0.8685824394226074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,4,1,balanced,0.033301333586374916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,4,1,power_law_1.01,0.36966400146484374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,4,1,power_law_1.01,0.38200960159301756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,4,1,power_law_1.01,0.16979199647903442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,4,1,power_law_1.01,0.3954751968383789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,4,1,power_law_1.01,0.17594879865646362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,4,1,power_law_1.01,0.43544321060180663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,4,1,power_law_1.01,0.4816768169403076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,4,1,power_law_1.01,0.5339136123657227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,4,power_law_1.01,14.946003723144532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,4,1,power_law_1.01,0.07463039755821228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,4,1,power_law_1.01,0.09037439823150635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,4,1,power_law_1.01,0.09286400079727172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,4,1,power_law_1.01,0.18250880241394044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,4,1,power_law_1.01,0.16019200086593627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,4,1,power_law_1.01,0.19148160219192506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,4,1,power_law_1.01,0.6706495761871338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,4,1,power_law_1.01,0.23400959968566895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,4,1,power_law_1.01,0.26472959518432615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,4,1,power_law_1.01,0.8147968292236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,4,1,balanced,0.055573334296544395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,4,1,power_law_1.01,0.2764672040939331
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,4,1,balanced,0.07858666777610779
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,4,1,balanced,0.07887466748555501
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,4,1,balanced,0.07947733501593272
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,4,1,balanced,0.08193600177764893
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,4,1,balanced,0.08753066261609395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,4,1,balanced,0.09305066863695781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,4,1,balanced,0.09335466225941975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,4,1,balanced,0.09231467048327129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,4,1,balanced,0.09368000427881877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,4,1,balanced,0.09936533371607463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,4,1,balanced,0.10014933347702026
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,4,1,balanced,0.1042080024878184
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,4,1,balanced,0.14998400211334229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,4,1,balanced,0.13133333126703897
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,4,1,balanced,0.11780800422032674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,4,1,balanced,0.23523199558258057
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,4,1,balanced,0.14706666270891824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,4,1,balanced,0.1953493356704712
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,4,1,balanced,0.24228266874949136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,4,1,balanced,0.34200000762939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,4,1,power_law_1.01,0.2842623949050903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,4,1,balanced,0.4430026610692342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,4,1,power_law_1.01,0.296012806892395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,4,1,balanced,0.6444533268610636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,4,1,power_law_1.01,0.32798080444335936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,4,1,balanced,0.8460640112559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,4,1,power_law_1.01,0.36197121143341066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,4,1,power_law_1.01,0.1611456036567688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,4,1,power_law_1.01,0.4052864074707031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,4,power_law_1.2,23.353543090820313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,4,1,power_law_1.01,0.4629824161529541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,4,1,power_law_1.01,1.1193984031677247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,4,power_law_1.2,5.333369445800781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,4,1,power_law_1.01,0.16212480068206786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,4,1,power_law_1.01,0.16522879600524903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,4,1,power_law_1.01,0.16574079990386964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,4,1,power_law_1.01,0.16902400255203248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,4,1,power_law_1.01,0.16929919719696046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,4,1,power_law_1.01,0.19080959558486937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,4,1,power_law_1.01,0.2036736011505127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,4,1,balanced,1.2422613302866619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,4,1,power_law_1.01,0.20861439704895018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,4,1,power_law_1.01,0.22864000797271727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,4,1,balanced,1.645045280456543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,4,1,power_law_1.01,0.27013120651245115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,4,1,power_law_1.01,0.2823231935501099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,4,1,power_law_1.01,0.32748799324035643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,4,1,power_law_1.01,0.42194561958312987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,4,1,power_law_1.01,0.618227195739746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,4,power_law_1.2,1.4844799995422364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,4,1,power_law_1.01,1.4217984199523925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,4,1,power_law_1.01,0.7627071857452392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,4,1,power_law_1.01,0.5165503978729248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,4,1,power_law_1.01,0.7265535831451416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,4,power_law_1.2,12.539718627929688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,4,1,power_law_1.01,0.9274944305419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,4,1,power_law_1.01,0.0371071994304657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,4,1,power_law_1.01,1.0744511604309082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,4,1,power_law_1.01,0.047603198885917665
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,4,1,power_law_1.01,0.06061440110206604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,4,1,power_law_1.01,2.06695671081543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,4,1,power_law_1.01,0.07589759826660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,4,1,power_law_1.01,0.07740160226821899
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,4,1,power_law_1.01,0.08172159790992736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,4,1,power_law_1.01,0.08747519850730896
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,4,1,power_law_1.01,0.09848319888114929
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,4,1,power_law_1.01,0.09100800156593322
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,4,1,power_law_1.01,0.0936191976070404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,4,1,power_law_1.01,0.10101759433746338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,4,1,power_law_1.01,0.09740800261497498
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,4,1,power_law_1.01,0.1056447982788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,4,1,power_law_1.01,0.11575679779052735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,4,1,power_law_1.01,0.11344640254974366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,4,1,power_law_1.01,0.12042239904403687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,4,1,power_law_1.01,1.3720512390136719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,4,1,power_law_1.01,0.13719040155410767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,4,1,power_law_1.01,1.3434176445007324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,4,1,power_law_1.01,0.16539520025253296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,4,1,power_law_1.01,2.6789567947387694
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,4,1,power_law_1.01,0.2125823974609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,4,1,power_law_1.01,1.9895744323730469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,4,1,power_law_1.01,0.21875200271606446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,4,1,power_law_1.01,1.7458879470825195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,4,1,power_law_1.01,0.28124799728393557
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,4,power_law_1.2,1.8718656539916991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,4,1,power_law_1.01,2.60699520111084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,4,1,power_law_1.01,0.3729664087295532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,4,1,power_law_1.01,2.5740928649902344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,4,1,power_law_1.01,4.0496257781982425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,4,power_law_1.2,6.855622100830078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,4,1,power_law_1.01,0.47170557975769045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,4,power_law_1.2,40.96831359863281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,4,1,power_law_1.2,0.09664000272750854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,4,1,power_law_1.2,0.11725440025329589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,4,1,power_law_1.2,0.14904320240020752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,4,1,power_law_1.2,0.25340158939361573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,4,1,power_law_1.2,0.2508224010467529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,4,1,power_law_1.2,0.2518656015396118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,4,1,power_law_1.2,0.25566079616546633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,4,1,power_law_1.2,0.2546367883682251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,4,1,power_law_1.2,0.2776959896087646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,4,1,power_law_1.2,0.30451838970184325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,4,1,power_law_1.01,3.837433624267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,4,1,power_law_1.2,0.3224128007888794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,4,1,power_law_1.2,0.3741631984710693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,4,1,power_law_1.2,0.37643520832061766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,4,1,power_law_1.2,0.4026175975799561
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,4,1,power_law_1.01,0.675603199005127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,4,1,power_law_1.2,0.42768001556396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,4,1,power_law_1.2,0.45484161376953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,4,1,power_law_1.2,0.5326975822448731
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,4,power_law_1.2,3.265024185180664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,4,1,power_law_1.2,0.6661888122558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,4,1,power_law_1.01,3.379635238647461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,4,1,power_law_1.2,0.830624008178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,4,1,power_law_1.01,0.8748479843139648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,4,1,power_law_1.2,1.1048704147338868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,4,1,power_law_1.01,5.02940788269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,4,1,power_law_1.2,1.4353983879089356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,4,1,power_law_1.2,2.05916805267334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,4,1,power_law_1.01,5.605503845214844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,4,1,power_law_1.01,5.050131225585938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,4,power_law_1.2,11.454220581054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,4,1,power_law_1.2,2.6926528930664064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,4,1,power_law_1.01,8.541165161132813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,4,1,power_law_1.2,3.9441600799560548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,4,power_law_1.2,23.857472229003907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,4,1,power_law_1.01,6.665055847167968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,4,1,power_law_1.2,0.08474239706993103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,4,1,power_law_1.2,0.09699199795722961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,4,1,power_law_1.2,0.11871999502182007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,4,1,power_law_1.2,0.1769215941429138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,4,1,power_law_1.2,0.16901760101318358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,4,1,power_law_1.2,0.17529599666595458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,4,1,power_law_1.2,0.1818112015724182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,4,1,power_law_1.2,0.19126399755477905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,4,1,power_law_1.2,0.20954880714416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,4,1,power_law_1.2,0.23875200748443604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,4,1,power_law_1.2,0.27073280811309813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,4,1,power_law_1.2,0.28356480598449707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,4,1,power_law_1.2,0.29492480754852296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,4,1,power_law_1.2,0.32394239902496336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,4,1,power_law_1.2,0.35843839645385744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,4,1,power_law_1.01,7.50948486328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,4,1,power_law_1.2,0.390828800201416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,4,1,power_law_1.2,5.618175888061524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,4,1,power_law_1.2,0.469868803024292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,4,1,power_law_1.2,0.6215744018554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,4,1,power_law_1.01,1.2640768051147462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,4,1,power_law_1.2,0.7732992172241211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,4,1,power_law_1.2,1.0797056198120116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,4,power_law_1.2,3.7074817657470702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,4,1,power_law_1.2,0.09050880074501037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,4,1,power_law_1.2,0.09738240242004395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,4,1,power_law_1.2,0.10309760570526123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,4,1,power_law_1.2,0.1711743950843811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,4,1,power_law_1.2,0.16416640281677247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,4,1,power_law_1.2,0.16430720090866088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,4,1,power_law_1.2,0.16621439456939696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,4,1,power_law_1.2,0.16627199649810792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,4,1,power_law_1.2,1.3763903617858886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,4,1,power_law_1.2,0.16946560144424438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,4,1,power_law_1.2,0.16871039867401122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,4,1,power_law_1.2,0.18354560136795045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,4,1,power_law_1.2,0.2027008056640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,4,1,power_law_1.2,0.20752639770507814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,4,1,power_law_1.2,0.2246335983276367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,4,1,power_law_1.2,0.26776320934295655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,4,1,power_law_1.2,0.2817215919494629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,4,1,power_law_1.2,0.3261823892593384
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,4,1,power_law_1.01,1.724985694885254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,4,1,power_law_1.2,0.4291071891784668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,4,1,power_law_1.2,0.5131392002105712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,4,1,power_law_1.01,11.960063934326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,4,1,power_law_1.2,1.9774463653564454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,4,balanced,0.09967999656995137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,4,balanced,0.11708799997965495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,4,balanced,0.1816533406575521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,4,balanced,0.30777599414189655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,4,balanced,0.5260213216145834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,4,1,power_law_1.2,0.7215487957000732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,4,balanced,0.9618986447652181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,4,balanced,0.9639306863149008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,4,balanced,0.9638026555379232
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,4,1,power_law_1.2,0.03758719861507416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,4,1,power_law_1.2,0.05118719935417175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,4,1,power_law_1.2,0.05406079888343811
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,4,1,power_law_1.2,0.07487360239028931
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,4,1,power_law_1.2,0.07756159901618957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,4,1,power_law_1.2,0.08155519962310791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,4,1,power_law_1.2,0.08789119720458985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,4,1,power_law_1.2,0.1003775954246521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,4,1,power_law_1.2,0.09159680008888245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,4,1,power_law_1.2,0.926848030090332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,4,1,power_law_1.2,0.09535999894142151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,4,balanced,1.1211466789245605
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,4,1,power_law_1.2,0.10154240131378174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,4,balanced,1.12391996383667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,4,1,power_law_1.2,0.10040960311889649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,4,balanced,1.1329546769460042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,4,1,power_law_1.2,0.10576640367507935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,4,balanced,1.1407946745554607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,4,balanced,1.155450661977132
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,4,1,power_law_1.2,0.11747200489044189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,4,balanced,1.1678400039672852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,4,1,power_law_1.2,0.11196160316467285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,4,balanced,1.1853013038635254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,4,1,power_law_1.2,1.3287039756774903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,4,balanced,1.203701337178548
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,4,1,power_law_1.2,0.12337919473648071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,4,balanced,1.236554702123006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,4,balanced,1.2982772986094158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,4,balanced,1.3628479639689128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,4,1,power_law_1.2,0.13428479433059692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,4,balanced,1.5036053657531738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,4,balanced,1.6702027320861816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,4,1,power_law_1.2,2.6009408950805666
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,4,1,power_law_1.2,0.1654911994934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,4,balanced,2.3703360557556152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,4,balanced,2.6793174743652344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,4,1,power_law_1.2,0.1721984028816223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,4,balanced,3.8378079732259116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,4,balanced,5.033919970194499
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,4,1,power_law_1.2,0.2287168025970459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,4,balanced,8.428799947102865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,4,1,power_law_1.2,0.2882368087768555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,4,1,power_law_1.2,8.618048095703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,4,power_law_1.2,16.646322631835936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,4,balanced,0.07660266757011414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,4,balanced,0.09428800145785014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,4,balanced,0.13379200299580893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,4,balanced,0.20315200090408325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,4,balanced,0.33352001508076984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,4,balanced,12.853407541910807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,4,balanced,0.632202665011088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,4,balanced,0.6365760167439779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,4,balanced,0.6399840116500854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,4,balanced,0.7530720233917236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,4,balanced,0.7587359746297201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,4,balanced,0.7729600270589193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,4,balanced,0.08014399806658427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,4,balanced,0.09319999814033508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,4,balanced,0.7808319727579752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,4,balanced,0.12032000223795573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,4,balanced,0.1893493334452311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,4,balanced,0.7923466364542643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,4,balanced,0.35093335310618085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,4,balanced,0.8145013650258383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,4,balanced,0.6522560119628906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,4,1,power_law_1.2,1.7430784225463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,4,balanced,0.8364319801330566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,4,balanced,0.6593173344930013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,4,balanced,0.6631840070088705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,4,balanced,0.6676639715830485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,4,balanced,0.6746346950531006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,4,balanced,0.6830080350240072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,4,1,power_law_1.01,9.969554901123047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,4,balanced,0.6930987040201823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,4,balanced,0.7035733064015707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,4,balanced,0.7234026590983073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,4,balanced,0.7425813674926758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,4,balanced,0.7605013052622477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,4,balanced,0.8594079812367758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,4,balanced,0.7953813076019287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,4,balanced,0.8977920214335123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,4,power_law_1.2,32.569784545898436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,4,balanced,0.8648373285929362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,4,balanced,0.9896852970123291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,4,balanced,0.9299733638763428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,4,balanced,1.0681227048238118
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,4,1,power_law_1.2,0.36560640335083006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,4,balanced,1.25436798731486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,4,balanced,1.4368054072062175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,4,balanced,2.1575093269348145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,4,1,power_law_1.2,3.8238719940185546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,4,power_law_1.01,0.15145599842071533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,4,balanced,2.5331145922342935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,4,balanced,1.0624852975209553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,4,balanced,1.2379679679870605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,4,balanced,3.653850555419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,4,balanced,1.7599679629007976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,4,balanced,2.071616013844808
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,4,balanced,0.03143999973932902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,4,power_law_1.01,0.30714879035949705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,4,balanced,0.04158399999141693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,4,balanced,0.06201600035031637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,4,balanced,0.09185066819190979
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,4,balanced,0.144405335187912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,4,balanced,2.9225387573242188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,4,balanced,0.2523893316586812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,4,balanced,0.258730669816335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,4,balanced,0.26288533210754395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,4,balanced,0.2648640076319377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,4,balanced,0.26842133204142254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,4,balanced,0.27456533908843994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,4,balanced,0.2682986656824748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,4,balanced,3.7886133193969727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,4,balanced,0.2751573324203491
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,4,balanced,0.2813653349876404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,4,balanced,0.2797013322512309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,4,balanced,4.748341242472331
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,4,balanced,0.28546667098999023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,4,balanced,0.29709867636362713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,4,balanced,0.34251733620961505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,4,balanced,0.32913599411646527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,4,balanced,0.5700106620788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,4,balanced,5.561808268229167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,4,balanced,0.40582935015360516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,4,balanced,0.41784000396728516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,4,balanced,0.4673226674397786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,4,balanced,6.973072052001953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,4,power_law_1.01,0.37297921180725097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,4,power_law_1.01,0.4905856132507324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,4,balanced,7.293274561564128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,4,power_law_1.01,0.5242432117462158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,4,power_law_1.01,0.11802879571914673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,4,power_law_1.01,0.6740223884582519
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,4,balanced,0.6738399664560953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,4,power_law_1.01,0.1965376019477844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,4,balanced,0.7677493095397949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,4,balanced,9.504554748535156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,4,power_law_1.01,0.7849215984344482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,4,power_law_1.01,0.23763840198516845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,4,balanced,1.0973973274230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,4,power_law_1.01,0.2864448070526123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,4,power_law_1.01,0.8576448440551758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,4,balanced,1.4321759541829426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,4,power_law_1.01,0.3765631914138794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,4,power_law_1.01,1.0327872276306151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,4,power_law_1.01,0.41597437858581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,4,power_law_1.01,0.10844800472259522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,4,power_law_1.01,1.0147071838378907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,4,power_law_1.01,0.5601408004760742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,4,power_law_1.01,0.18220160007476807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,4,power_law_1.01,0.5067264080047608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,4,power_law_1.01,0.9780735969543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,4,power_law_1.01,0.25873920917510984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,4,power_law_1.01,0.7097087860107422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,4,power_law_1.01,0.3160959959030151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,4,power_law_1.01,0.6942848205566406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,4,power_law_1.01,0.4085696220397949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,4,power_law_1.01,0.7047103881835938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,4,power_law_1.01,0.49954562187194823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,4,power_law_1.01,0.7334015846252442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,4,power_law_1.01,0.5552768230438232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,4,1,power_law_1.2,2.5542272567749023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,4,power_law_1.01,0.7423232078552247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,4,power_law_1.01,0.5632383823394775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,4,power_law_1.01,1.0226048469543456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,4,1,power_law_1.2,0.48601598739624025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,4,power_law_1.01,0.5634751796722413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,4,power_law_1.01,1.0694144248962403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,4,power_law_1.01,0.5777728080749511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,4,power_law_1.01,0.6018688201904296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,4,power_law_1.01,1.087564754486084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,4,power_law_1.01,0.5904640197753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,4,power_law_1.01,1.12609920501709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,4,power_law_1.01,0.6042816162109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,4,power_law_1.01,0.7673791885375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,4,power_law_1.01,1.1574272155761718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,4,power_law_1.01,0.05742719769477844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,4,power_law_1.01,0.8164223670959473
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,4,power_law_1.01,0.08380799889564514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,4,power_law_1.01,0.8379136085510254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,4,power_law_1.01,0.1078976035118103
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,4,power_law_1.01,0.12881280183792115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,4,power_law_1.01,0.9096639633178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,4,power_law_1.01,0.15635199546813966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,4,power_law_1.01,1.0359423637390137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,4,power_law_1.01,0.6504767894744873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,4,power_law_1.01,0.19347200393676758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,4,power_law_1.01,0.21639680862426758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,4,power_law_1.01,0.6703423976898193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,4,power_law_1.01,1.2106623649597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,4,power_law_1.01,1.2281023979187011
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,4,power_law_1.01,0.22104959487915038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,4,power_law_1.01,0.7229184150695801
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,4,power_law_1.01,0.21811199188232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,4,power_law_1.01,1.3351743698120118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,4,power_law_1.01,1.5741503715515137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,4,power_law_1.01,0.7559999942779541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,4,power_law_1.01,0.8613759994506835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,4,power_law_1.01,1.488652801513672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,4,power_law_1.01,1.7822015762329102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,4,power_law_1.01,0.9857983589172363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,4,power_law_1.01,1.7273216247558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,4,power_law_1.01,1.2202431678771972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,4,power_law_1.01,2.461631965637207
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,4,power_law_1.01,0.24225919246673583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,4,power_law_1.01,2.0708160400390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,4,power_law_1.01,1.432863998413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,4,power_law_1.01,0.23554561138153077
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,4,power_law_1.01,0.2342463970184326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,4,power_law_1.01,3.082592010498047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,4,power_law_1.01,0.24896640777587892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,4,power_law_1.01,1.9100223541259767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,4,power_law_1.01,2.6790655136108397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,4,power_law_1.01,0.25262720584869386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,4,1,power_law_1.2,0.6689727783203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,4,power_law_1.01,0.26063361167907717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,4,power_law_1.01,0.270361590385437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,4,power_law_1.01,2.398534393310547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,4,power_law_1.01,0.286572790145874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,4,power_law_1.01,3.408428955078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,4,power_law_1.01,4.51357421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,4,power_law_1.01,0.30894720554351807
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,4,power_law_1.01,0.3295167922973633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,4,power_law_1.01,0.48853120803833006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,4,power_law_1.01,3.475948715209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,4,1,power_law_1.2,5.045049667358398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,4,power_law_1.01,0.41294078826904296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,4,power_law_1.01,4.605279922485352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,4,power_law_1.01,0.551423978805542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,4,power_law_1.01,5.507276916503907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,4,power_law_1.01,0.6092095851898194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,4,power_law_1.01,4.397151947021484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,4,power_law_1.01,0.7567808151245117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,4,power_law_1.01,0.9987456321716308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,4,power_law_1.01,5.98430061340332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,4,power_law_1.01,1.3641152381896973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,4,power_law_1.01,6.297209548950195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,4,power_law_1.01,8.457453155517578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,4,power_law_1.01,2.2592512130737306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,4,power_law_1.2,0.15210239887237548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,4,power_law_1.2,0.2667327880859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,4,power_law_1.2,0.32750720977783204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,4,power_law_1.2,0.43258237838745117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,4,power_law_1.01,9.551193237304688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,4,power_law_1.2,0.5581567764282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,4,power_law_1.2,0.6093376159667969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,4,power_law_1.01,8.205919647216797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,4,power_law_1.2,0.8351167678833008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,4,power_law_1.2,0.11704319715499878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,4,power_law_1.01,10.341964721679688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,4,power_law_1.2,0.19653120040893554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,4,power_law_1.2,0.8038847923278809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,4,power_law_1.2,0.21827199459075927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,4,power_law_1.2,0.9887871742248535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,4,power_law_1.2,0.2665855884552002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,4,power_law_1.2,0.367193603515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,4,power_law_1.2,0.977683162689209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,4,power_law_1.2,0.10835200548171997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,4,power_law_1.2,0.4511871814727783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,4,power_law_1.2,0.9919872283935547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,4,power_law_1.2,0.1795199990272522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,4,power_law_1.2,0.5170048236846924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,4,power_law_1.01,11.3731201171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,4,power_law_1.2,0.1787775993347168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,4,power_law_1.2,1.0844863891601562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,4,power_law_1.2,0.5011519908905029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,4,power_law_1.2,0.28853759765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,4,power_law_1.2,1.0735103607177734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,4,power_law_1.2,0.3889024019241333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,4,power_law_1.2,1.108339214324951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,4,power_law_1.2,1.1372223854064942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,4,power_law_1.2,1.1645567893981934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,4,power_law_1.2,0.6817855834960938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,4,power_law_1.2,1.2742079734802245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,4,power_law_1.2,0.6978496074676513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,4,power_law_1.2,0.45384960174560546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,4,power_law_1.2,1.3478015899658202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,4,power_law_1.2,0.6499648094177246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,4,power_law_1.2,0.5261055946350097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,4,power_law_1.2,0.7453760147094727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,4,power_law_1.2,0.523686408996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,4,power_law_1.2,1.5495360374450684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,4,power_law_1.2,0.733132791519165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,4,power_law_1.2,0.5901055812835694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,4,power_law_1.2,0.7836095809936523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,4,power_law_1.2,0.5633664131164551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,4,power_law_1.2,1.7673023223876954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,4,power_law_1.2,0.7994624137878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,4,power_law_1.2,0.604262399673462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,4,power_law_1.2,0.8607168197631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,4,power_law_1.2,0.6450623989105224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,4,power_law_1.2,2.1457151412963866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,4,power_law_1.2,0.6400256156921387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,4,power_law_1.2,0.9635392189025879
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,4,power_law_1.2,0.05578879714012146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,4,power_law_1.2,0.696281623840332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,4,power_law_1.2,0.08284800052642823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,4,power_law_1.2,1.0953856468200684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,4,power_law_1.2,2.6594751358032225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,4,power_law_1.2,0.7101056098937988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,4,power_law_1.2,0.0920960009098053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,4,power_law_1.2,1.2606911659240723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,4,power_law_1.2,0.728934383392334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,4,power_law_1.2,0.7649663925170899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,4,power_law_1.2,3.4698368072509767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,4,power_law_1.2,1.5870016098022461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,4,power_law_1.2,0.8960639953613281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,4,power_law_1.2,1.9598400115966796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,4,power_law_1.2,1.0346752166748048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,4,1,power_law_1.2,3.3857471466064455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,4,power_law_1.2,0.1261888027191162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,4,power_law_1.2,1.26112642288208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,4,power_law_1.2,0.16158080101013184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,4,power_law_1.2,5.317529678344727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,4,power_law_1.2,2.656083106994629
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,4,power_law_1.2,0.16845439672470092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,4,power_law_1.2,1.5120767593383788
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,4,power_law_1.2,0.22288000583648682
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,4,power_law_1.2,0.22118399143218995
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,4,power_law_1.2,0.23192319869995118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,4,power_law_1.2,2.0416831970214844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,4,power_law_1.2,0.23006720542907716
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,4,power_law_1.2,0.2466304063796997
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,4,power_law_1.2,0.2420351982116699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,4,power_law_1.2,6.4279228210449215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,4,power_law_1.2,2.478329658508301
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,4,power_law_1.2,0.2563647985458374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,4,power_law_1.2,0.25963521003723145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,4,power_law_1.2,0.2500799894332886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,4,power_law_1.2,3.1662464141845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,4,power_law_1.2,0.2752959966659546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,4,power_law_1.2,0.29569919109344484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,4,power_law_1.2,3.521708679199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,4,power_law_1.2,0.3228480100631714
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,4,power_law_1.2,0.35113599300384524
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,4,power_law_1.2,0.44379520416259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,4,power_law_1.2,4.3222911834716795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,4,power_law_1.2,0.4683519840240479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,4,power_law_1.2,9.947090911865235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,4,power_law_1.2,5.045145416259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,4,power_law_1.2,0.5827968120574951
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,4,power_law_1.2,0.6576640129089355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,4,power_law_1.2,0.8862848281860352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,4,power_law_1.2,6.231852722167969
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,4,power_law_1.2,1.1011967658996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,4,1,power_law_1.2,0.8686079978942871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,4,power_law_1.2,6.964991760253906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,4,power_law_1.2,1.5079872131347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,4,power_law_1.2,2.590675163269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,4,1,balanced,0.07772266864776611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,4,power_law_1.2,12.269023895263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,4,power_law_1.2,8.371488189697265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,4,1,balanced,0.07109866539637248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,4,1,balanced,0.08851200342178345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,4,1,balanced,0.11597866813341777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,4,1,balanced,0.20688533782958984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,4,1,balanced,0.3535573482513428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,4,1,balanced,0.6695360342661539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,4,1,balanced,0.6696159839630127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,4,1,balanced,0.6716533501942953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,4,power_law_1.2,8.954688262939452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,4,1,balanced,0.6702346801757812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,4,1,balanced,0.672661304473877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,4,1,balanced,0.09816533327102661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,4,1,balanced,0.14247467120488486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,4,1,balanced,0.29766400655110675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,4,1,balanced,0.5362293322881063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,4,1,balanced,0.9663893381754557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,4,1,balanced,0.9667147000630697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,4,1,balanced,0.058037335673967995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,4,1,balanced,0.9665599664052328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,4,1,power_law_1.2,7.482176208496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,4,1,balanced,0.678213357925415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,4,1,balanced,0.683631976445516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,4,1,balanced,0.6887839635213217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,4,1,balanced,0.69651198387146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,4,1,balanced,0.7154133319854736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,4,1,balanced,0.7256426811218262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,4,1,balanced,0.7533386548360189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,4,1,balanced,0.9684800306955973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,4,1,balanced,0.8113280137379965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,4,1,power_law_1.2,11.986316680908203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,4,1,balanced,0.9701386292775472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,4,power_law_1.2,12.112608337402344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,4,1,balanced,0.9729546705881754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,4,1,balanced,0.9712479909261068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,4,1,balanced,0.975056012471517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,4,1,balanced,0.9840853214263916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,4,1,balanced,1.054357369740804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,4,1,balanced,1.064522663752238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,4,1,balanced,0.8673813343048096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,4,1,balanced,0.9988746643066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,4,1,power_law_1.01,0.07686399817466735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,4,1,balanced,0.0763679991165797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,4,1,balanced,0.09847467144330342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,4,1,balanced,1.135258674621582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,4,1,power_law_1.01,0.09583359956741333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,4,1,power_law_1.01,0.13950719833374023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,4,1,balanced,1.681440035502116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,4,1,power_law_1.01,0.28155519962310793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,4,1,balanced,1.952309290568034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,4,1,power_law_1.01,0.3953344106674194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,4,1,balanced,1.0864746570587158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,4,1,power_law_1.01,0.5242623805999755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,4,1,balanced,1.1309813658396404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,4,1,balanced,2.7736587524414062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,4,1,balanced,0.1518239974975586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,4,1,power_law_1.01,0.7243135929107666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,4,1,balanced,1.1765920321146648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,4,1,balanced,1.286911964416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,4,1,power_law_1.01,0.7445888042449951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,4,1,balanced,3.5995893478393555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,4,1,balanced,0.02923733244339625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,4,1,balanced,0.038831998904546104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,4,1,balanced,0.058149332801500954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,4,1,balanced,1.4074187278747559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,4,1,power_law_1.01,0.7671743869781494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,4,1,balanced,0.0867733359336853
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,4,1,balanced,0.14226133624712625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,4,1,balanced,0.25200533866882324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,4,1,power_law_1.01,0.7820032119750977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,4,1,balanced,1.9877920150756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,4,1,power_law_1.01,0.8228351593017578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,4,1,balanced,5.269952138264974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,4,1,balanced,2.2202399571736655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,4,1,power_law_1.01,0.8554240226745605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,4,1,power_law_1.01,0.8978367805480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,4,1,balanced,3.147461255391439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,4,1,power_law_1.01,0.9515263557434082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,4,1,balanced,0.2577706575393677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,4,1,balanced,0.260858674844106
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,4,1,balanced,0.2640213370323181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,4,1,balanced,6.9377492268880205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,4,1,balanced,0.26769065856933594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,4,1,power_law_1.01,0.9942336082458496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,4,1,balanced,0.27324267228444415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,4,1,balanced,4.075306574503581
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,4,1,balanced,0.27774399518966675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,4,1,power_law_1.01,0.0649728000164032
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,4,1,balanced,0.28520532449086505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,4,1,power_law_1.01,1.0134464263916017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,4,1,balanced,0.29467199246088666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,4,1,power_law_1.01,0.08436480164527893
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,4,1,balanced,0.32018133004506427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,4,1,balanced,0.3317333261171977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,4,1,power_law_1.01,1.0699263572692872
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,4,1,balanced,0.3524640003840129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,4,1,balanced,0.3629973332087199
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,4,1,balanced,0.3909653425216675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,4,1,power_law_1.01,1.174790382385254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,4,1,balanced,0.429418683052063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,4,1,balanced,5.989429473876953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,4,1,balanced,0.2810293237368266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,4,1,balanced,0.4718666474024455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,4,1,balanced,0.5217440128326416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,4,1,power_law_1.01,1.2408767700195313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,4,1,power_law_1.01,1.4195072174072265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,4,1,power_law_1.01,0.11121280193328857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,4,1,power_law_1.01,0.20049281120300294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,4,1,power_law_1.01,1.6262016296386719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,4,1,power_law_1.01,0.2935935974121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,4,1,balanced,0.5674346685409546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,4,1,power_law_1.01,0.37104001045227053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,4,1,balanced,9.763578414916992
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,4,1,balanced,0.6128799915313721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,4,1,power_law_1.01,2.0539199829101564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,4,1,power_law_1.01,0.4927040100097656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,4,1,balanced,1.083136002222697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,4,1,power_law_1.01,0.5231743812561035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,4,1,balanced,0.9164640108744303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,4,1,power_law_1.01,0.5393407821655274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,4,1,power_law_1.01,2.44901123046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,4,1,balanced,1.300160010655721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,4,1,power_law_1.01,0.5441408157348633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,4,1,power_law_1.01,0.05674239993095398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,4,1,balanced,1.689242680867513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,4,1,power_law_1.01,0.594982385635376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,4,1,power_law_1.01,0.07486079931259156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,4,1,power_law_1.01,0.6414400100708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,4,1,power_law_1.01,3.3252799987792967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,4,1,power_law_1.01,0.6535935878753663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,4,1,power_law_1.2,1.2886528015136718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,4,1,power_law_1.01,0.743065595626831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,4,1,balanced,0.5423093239466349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,4,1,power_law_1.01,4.2875007629394535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,4,1,power_law_1.01,0.6932544231414794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,4,1,power_law_1.01,0.09667840003967285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,4,1,power_law_1.01,0.1486207962036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,4,1,power_law_1.01,0.7082111835479736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,4,1,power_law_1.01,0.24970879554748535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,4,1,power_law_1.01,0.746943998336792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,4,1,power_law_1.01,0.2859839916229248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,4,1,power_law_1.01,0.8568896293640137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,4,1,power_law_1.01,0.4027520179748535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,4,1,power_law_1.01,5.853311920166016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,4,1,power_law_1.01,0.9531647682189941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,4,1,power_law_1.01,0.4139455795288086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,4,1,power_law_1.01,0.02752000093460083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,4,1,power_law_1.01,0.4272575855255127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,4,1,power_law_1.01,1.141932773590088
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,4,1,power_law_1.01,0.035820800065994265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,4,1,power_law_1.01,0.43732481002807616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,4,1,power_law_1.01,0.05611519813537598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,4,1,power_law_1.01,1.3497471809387207
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,4,1,power_law_1.01,0.08516479730606079
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,4,1,power_law_1.01,0.11715199947357177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,4,1,power_law_1.01,1.751372718811035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,4,1,power_law_1.01,0.15731199979782104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,4,1,power_law_1.01,0.20515201091766358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,4,1,power_law_1.01,2.172332763671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,4,1,power_law_1.01,0.21587200164794923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,4,1,balanced,0.523253321647644
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,4,1,power_law_1.01,0.2187391996383667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,4,1,balanced,0.526965339978536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,4,1,power_law_1.01,0.22738559246063234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,4,1,balanced,0.5293600161870321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,4,1,balanced,0.5436373154322306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,4,1,power_law_1.01,0.2363840103149414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,4,1,balanced,0.5505866607030233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,4,1,power_law_1.01,0.24433279037475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,4,1,balanced,0.5544853210449219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,4,1,balanced,0.5523360172907511
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,4,1,power_law_1.01,0.2585279941558838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,4,1,power_law_1.01,8.02880630493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,4,1,balanced,0.5661120017369589
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,4,1,power_law_1.01,0.2756736040115356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,4,1,balanced,0.57259734471639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,4,1,power_law_1.2,0.07716479897499084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,4,1,power_law_1.01,0.2973952054977417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,4,1,balanced,0.6787786483764648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,4,1,power_law_1.01,0.4524672031402588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,4,1,power_law_1.2,0.09520639777183533
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,4,1,power_law_1.01,0.32325758934021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,4,1,power_law_1.2,0.13886719942092896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,4,1,power_law_1.01,0.483519983291626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,4,1,power_law_1.01,0.3728384017944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,4,1,power_law_1.2,0.2739648103713989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,4,1,power_law_1.01,3.0110912322998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,4,1,power_law_1.01,0.3605439901351929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,4,1,power_law_1.2,0.36455678939819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,4,1,power_law_1.01,0.4232319831848145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,4,1,power_law_1.2,0.49193601608276366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,4,1,power_law_1.01,0.4365824222564697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,4,1,power_law_1.01,0.5125311851501465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,4,1,power_law_1.2,0.702016019821167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,4,1,power_law_1.01,3.840262222290039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,4,1,power_law_1.01,0.5542079925537109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,4,1,power_law_1.2,0.7350336074829101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,4,1,power_law_1.01,0.503929615020752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,4,1,power_law_1.01,0.6676671981811524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,4,1,power_law_1.2,0.7446591854095459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,4,1,power_law_1.01,0.5111551761627198
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,4,1,power_law_1.01,0.8280575752258301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,4,1,power_law_1.2,0.7679808139801025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,4,1,power_law_1.01,0.5704063892364502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,4,1,power_law_1.01,0.5916927814483642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,4,1,power_law_1.2,0.8139840126037597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,4,1,power_law_1.01,5.511507034301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,4,1,balanced,0.7115786870320638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,4,1,power_law_1.2,0.8746879577636719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,4,1,balanced,0.8241120179494222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,4,1,balanced,0.9056320190429688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,4,1,power_law_1.2,0.9013824462890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,4,1,power_law_1.2,0.9822463989257812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,4,1,power_law_1.2,1.0074048042297363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,4,1,power_law_1.01,7.159474945068359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,4,1,power_law_1.2,0.06624640226364135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,4,1,power_law_1.2,0.08430079817771911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,4,1,power_law_1.2,0.1109120011329651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,4,1,power_law_1.2,0.19896960258483887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,4,1,power_law_1.01,1.0287103652954102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,4,1,power_law_1.2,0.2523008108139038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,4,1,power_law_1.01,0.7445248126983642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,4,1,power_law_1.2,0.3459968090057373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,4,1,power_law_1.01,1.4177663803100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,4,1,power_law_1.01,0.890777587890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,4,1,power_law_1.2,0.4896255970001221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,4,1,power_law_1.01,0.8901760101318359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,4,1,power_law_1.2,0.5100800037384033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,4,1,power_law_1.01,2.0235519409179688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,4,1,power_law_1.2,0.5349567890167236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,4,1,power_law_1.01,1.0654272079467773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,4,1,power_law_1.2,1.0191424369812012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,4,1,power_law_1.2,0.5612480163574218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,4,1,power_law_1.2,0.5989120006561279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,4,1,power_law_1.01,1.2291071891784668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,4,1,power_law_1.2,0.6339136123657226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,4,1,power_law_1.2,5.030316925048828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,4,1,power_law_1.2,0.05601919889450073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,4,1,power_law_1.2,0.6657599925994873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,4,1,power_law_1.2,0.7605631828308106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,4,1,power_law_1.2,0.6931968212127686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,4,1,power_law_1.2,1.1317119598388672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,4,1,power_law_1.2,0.7150720119476318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,4,1,power_law_1.2,0.7787456035614013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,4,1,power_law_1.01,1.6141632080078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,4,1,power_law_1.2,0.8803071975708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,4,1,balanced,1.0379040241241455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,4,1,power_law_1.2,0.9851072311401368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,4,1,power_law_1.01,1.9575424194335938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,4,1,balanced,1.5411094029744465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,4,1,power_law_1.2,1.1747584342956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,4,1,balanced,1.7061386108398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,4,1,power_law_1.2,1.1904895782470704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,4,1,power_law_1.01,2.684396743774414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,4,1,power_law_1.2,1.3778688430786132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,4,1,balanced,2.409776051839193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,4,1,power_law_1.2,1.2470911979675292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,4,1,power_law_1.2,1.8010175704956055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,4,1,power_law_1.2,0.07371519804000855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,4,1,balanced,3.131264050801595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,4,1,power_law_1.01,3.4173633575439455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,4,1,power_law_1.2,0.09566720128059387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,4,1,power_law_1.2,2.203289604187012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,4,1,power_law_1.2,0.1384063959121704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,4,1,balanced,4.605552037556966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,4,1,power_law_1.2,3.039193534851074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,4,1,power_law_1.01,4.864409637451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,4,1,power_law_1.2,1.4576448440551757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,4,1,power_law_1.2,1.6539392471313477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,4,1,balanced,6.0764109293619795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,4,1,power_law_1.2,3.8768959045410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,4,1,power_law_1.2,9.945772552490235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,4,balanced,0.08323200047016144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,4,balanced,0.09682666261990865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,4,1,power_law_1.2,2.073017692565918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,4,1,power_law_1.01,6.310393524169922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,4,balanced,0.051738664507865906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,4,balanced,0.08556266625722249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,4,balanced,0.12153599659601848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,4,balanced,0.17791465918223062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,4,1,power_law_1.2,5.5435840606689455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,4,balanced,0.13917866349220276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,4,balanced,0.26576000452041626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,4,1,power_law_1.2,0.22879359722137452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,4,balanced,0.4656906525293986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,4,balanced,0.8260800043741862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,4,1,power_law_1.2,0.26582400798797606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,4,balanced,0.8317013581593832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,4,1,power_law_1.2,2.480601692199707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,4,balanced,0.833184003829956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,4,balanced,0.2909226616223653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,4,balanced,0.957530657450358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,4,balanced,0.5434879859288534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,4,balanced,0.9612267017364502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,4,balanced,0.548799991607666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,4,balanced,0.5501386721928915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,4,balanced,0.9730346997578939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,4,balanced,0.6561546723047892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,4,1,power_law_1.2,3.300729751586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,4,balanced,0.977125326792399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,4,balanced,0.6628746589024862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,4,balanced,0.9858720302581787
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,4,1,power_law_1.2,0.0297791987657547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,4,balanced,1.003061294555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,4,1,power_law_1.2,0.037452799081802365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,4,balanced,1.0173493226369221
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,4,1,power_law_1.2,0.05761280059814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,4,balanced,1.037893295288086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,4,1,power_law_1.2,7.2127937316894535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,4,1,power_law_1.2,4.319020843505859
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,4,1,power_law_1.2,0.08178560137748718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,4,balanced,1.0633227030436199
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,4,1,power_law_1.2,0.11004159450531006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,4,1,power_law_1.2,0.15141119956970214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,4,balanced,0.6746506690979004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,4,balanced,0.6847413380940756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,4,balanced,0.0724533349275589
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,4,1,power_law_1.2,0.19894399642944335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,4,balanced,0.08335466186205547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,4,balanced,0.10957866907119751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,4,balanced,0.16798933347066244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,4,1,power_law_1.2,0.21446399688720702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,4,balanced,0.30396799246470135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,4,balanced,0.5688746770222982
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,4,1,power_law_1.2,0.22188160419464112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,4,balanced,0.5755946636199951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,4,1,power_law_1.2,6.224518585205078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,4,1,power_law_1.2,0.39864320755004884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,4,1,power_law_1.2,0.2279871940612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,4,balanced,1.1226186752319336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,4,1,power_law_1.2,0.2403968095779419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,4,balanced,1.1888480186462402
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,4,1,power_law_1.2,0.2512063980102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,4,balanced,1.325482686360677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,4,1,power_law_1.2,0.2586560010910034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,4,balanced,0.6939626534779867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,4,1,power_law_1.2,0.28305280208587646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,4,balanced,1.4770506223042805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,4,1,power_law_1.2,0.30376319885253905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,4,1,power_law_1.2,0.32952959537506105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,4,1,power_law_1.2,0.3837759971618652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,4,1,power_law_1.2,8.207097625732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,4,balanced,0.7149653434753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,4,balanced,0.7356906731923422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,4,balanced,0.7548906803131104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,4,balanced,2.0812320709228516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,4,balanced,0.7910133202870687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,4,balanced,0.8761813640594482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,4,balanced,2.374239921569824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,4,balanced,0.9552000363667806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,4,balanced,1.1148426532745361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,4,balanced,1.2774133682250977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,4,balanced,3.396160125732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,4,1,power_law_1.2,0.40320000648498533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,4,balanced,0.5788960059483846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,4,balanced,1.9095253944396973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,4,balanced,0.5853653351465861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,4,1,power_law_1.2,0.4194943904876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,4,balanced,0.5881333351135254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,4,1,power_law_1.2,0.3721343994140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,4,1,power_law_1.2,0.42986240386962893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,4,balanced,4.512063980102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,4,1,power_law_1.2,0.4371840000152588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,4,1,power_law_1.2,0.4667647838592529
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,4,1,power_law_1.2,1.7073535919189453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,4,1,power_law_1.2,0.45102720260620116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,4,1,power_law_1.2,0.4882815837860107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,4,power_law_1.01,0.12748160362243652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,4,1,power_law_1.2,0.5055168151855469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,4,power_law_1.01,0.2555840015411377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,4,balanced,0.029445332785447437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,4,1,power_law_1.2,0.5257279872894287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,4,power_law_1.01,0.32896640300750735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,4,1,power_law_1.2,0.5757887840270997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,4,balanced,0.5990133285522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,4,power_law_1.01,0.4313471794128418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,4,balanced,0.6071946620941162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,4,1,power_law_1.2,0.6001279830932618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,4,balanced,0.6173173189163208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,4,power_law_1.01,0.4745471954345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,4,balanced,0.6341120004653931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,4,1,power_law_1.2,0.7725567817687988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,4,balanced,0.6535573403040568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,4,power_law_1.01,0.5864575862884521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,4,balanced,0.6700586477915446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,4,1,power_law_1.2,0.5292096138000488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,4,1,power_law_1.2,0.9233792304992676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,4,balanced,0.7029439608256022
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,4,balanced,0.03618666778008143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,4,balanced,0.7679573694864908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,4,1,power_law_1.2,0.9149503707885742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,4,balanced,7.457109451293945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,4,balanced,2.251749356587728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,4,1,power_law_1.2,1.0940352439880372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,4,balanced,3.2331838607788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,4,power_law_1.01,0.6468671798706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,4,1,power_law_1.2,0.5728640079498291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,4,balanced,0.05641599992911021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,4,balanced,0.08290133376916249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,4,power_law_1.01,0.6879487991333008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,4,balanced,0.8284640312194824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,4,balanced,4.204799969991048
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,4,1,power_law_1.2,0.6801983833312988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,4,balanced,0.9501012961069742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,4,power_law_1.01,0.8794431686401367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,4,balanced,10.248501459757486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,4,1,power_law_1.2,0.8460224151611329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,4,power_law_1.01,0.8739392280578613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,4,power_law_1.01,0.10613759756088256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,4,power_law_1.01,0.1772096037864685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,4,1,power_law_1.2,1.0485312461853027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,4,power_law_1.01,0.8463232040405273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,4,balanced,6.162933349609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,4,power_law_1.01,0.22074239253997802
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,4,balanced,0.12843199570973715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,4,power_law_1.01,0.9307456016540527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,4,balanced,0.21969600518544516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,4,power_law_1.01,0.26472959518432615
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,4,1,power_law_1.2,1.4335103988647462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,4,balanced,0.2247999906539917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,4,balanced,0.22804800669352213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,4,power_law_1.01,0.3436736106872559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,4,power_law_1.01,0.9089856147766113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,4,balanced,1.090901295344035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,4,power_law_1.01,0.38871040344238283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,4,power_law_1.01,0.9137791633605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,4,1,power_law_1.2,2.1098751068115233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,4,balanced,1.5736692746480305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,4,power_law_1.01,0.4648320198059082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,4,power_law_1.01,0.9352895736694335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,4,power_law_1.01,0.4561408042907715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,4,balanced,1.8475626309712727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,4,balanced,8.392112096150717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,4,power_law_1.01,0.9980223655700684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,4,power_law_1.01,0.577126407623291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,4,balanced,0.23224000136057535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,4,balanced,2.6325066884358725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,4,power_law_1.01,0.5599552154541015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,4,power_law_1.01,1.0490176200866699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,4,power_law_1.01,0.6290815830230713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,4,power_law_1.01,1.1843711853027343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,4,power_law_1.01,0.6310976028442383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,4,balanced,3.433055877685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,4,power_law_1.01,1.2249024391174317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,4,power_law_1.01,0.6506879806518555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,4,power_law_1.01,0.67674880027771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,4,power_law_1.01,1.5428288459777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,4,power_law_1.01,0.10040960311889649
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,4,balanced,0.2343626618385315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,4,power_law_1.01,0.7090559959411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,4,balanced,0.23948800563812256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,4,balanced,5.019482612609863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,4,balanced,0.2343626618385315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,4,power_law_1.01,0.15896960496902465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,4,balanced,0.23817066351572672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,4,power_law_1.01,0.7223104000091553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,4,power_law_1.01,1.789504051208496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,4,power_law_1.01,0.22174720764160155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,4,power_law_1.01,0.8091903686523437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,4,power_law_1.01,0.2730623960494995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,4,power_law_1.01,0.9398079872131347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,4,power_law_1.01,2.4787391662597655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,4,balanced,6.576976140340169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,4,power_law_1.01,1.0606911659240723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,4,power_law_1.2,0.12833280563354493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,4,balanced,0.24615466594696045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,4,power_law_1.01,1.3202303886413573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,4,power_law_1.01,2.985228729248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,4,power_law_1.2,0.2584383964538574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,4,power_law_1.01,0.051123201847076416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,4,1,power_law_1.2,1.269215965270996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,4,power_law_1.2,0.28181118965148927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,4,power_law_1.01,0.07567999958992004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,4,power_law_1.01,1.570911979675293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,4,power_law_1.01,0.3213248014450073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,4,power_law_1.2,0.35886080265045167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,4,power_law_1.01,0.09656320214271545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,4,1,power_law_1.2,1.6320768356323243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,4,power_law_1.01,0.40714240074157715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,4,power_law_1.2,0.46048641204833984
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,4,power_law_1.01,0.12135679721832275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,4,power_law_1.01,3.9125953674316407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,4,power_law_1.01,2.2132287979125977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,4,power_law_1.01,0.493452787399292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,4,power_law_1.2,0.5175487995147705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,4,power_law_1.01,0.14011520147323608
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,4,balanced,0.245253324508667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,4,power_law_1.01,0.48906240463256834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,4,power_law_1.01,0.16595200300216675
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,4,balanced,0.24851200977961221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,4,power_law_1.2,0.7495872020721436
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,4,balanced,0.25869866212209064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,4,power_law_1.01,0.19625600576400756
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,4,balanced,0.3012053370475769
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,4,balanced,0.28679466247558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,4,power_law_1.01,2.747782325744629
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,4,power_law_1.01,0.19406720399856567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,4,balanced,0.5046720107396444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,4,balanced,0.34970664978027344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,4,power_law_1.01,0.19368959665298463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,4,power_law_1.01,5.082566452026367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,4,balanced,0.36329599221547443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,4,power_law_1.01,0.1971392035484314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,4,power_law_1.01,0.21021440029144287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,4,1,power_law_1.2,2.005171203613281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,4,power_law_1.01,0.20887041091918945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,4,power_law_1.01,0.20890240669250487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,4,power_law_1.01,0.5176511764526367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,4,power_law_1.2,0.6934847831726074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,4,power_law_1.01,0.22355198860168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,4,power_law_1.01,0.5642623901367188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,4,power_law_1.01,0.22897279262542725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,4,1,power_law_1.2,6.670317077636719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,4,1,power_law_1.2,2.7257471084594727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,4,power_law_1.2,0.9156160354614258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,4,power_law_1.01,0.5089216232299805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,4,power_law_1.01,0.22978560924530028
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,4,balanced,0.4054933389027913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,4,power_law_1.2,0.10652159452438355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,4,power_law_1.2,0.7880896091461181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,4,power_law_1.01,7.595161437988281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,4,power_law_1.2,0.17562880516052246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,4,power_law_1.01,3.8356990814208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,4,power_law_1.2,0.8776448249816895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,4,power_law_1.2,0.18110719919204712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,4,1,power_law_1.2,3.4467136383056642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,4,power_law_1.2,0.21203839778900146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,4,power_law_1.2,0.9216511726379395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,4,power_law_1.2,0.3276479959487915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,4,power_law_1.2,0.9075776100158691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,4,power_law_1.2,0.3005120038986206
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,4,power_law_1.01,0.24428799152374267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,4,power_law_1.01,4.854694366455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,4,balanced,0.5854399998982748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,4,power_law_1.2,0.5140351772308349
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,4,power_law_1.01,0.2661119937896729
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,4,balanced,0.671445369720459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,4,1,power_law_1.2,4.900531387329101
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,4,power_law_1.01,0.2955712080001831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,4,balanced,0.9582986831665039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,4,power_law_1.01,0.34927361011505126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,4,balanced,1.2449920177459717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,4,power_law_1.01,0.3996543884277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,4,power_law_1.01,0.45862398147583006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,4,power_law_1.2,0.9318528175354004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,4,power_law_1.01,0.5298304080963134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,4,power_law_1.01,7.332704162597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,4,power_law_1.2,0.9583807945251465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,4,power_law_1.01,0.6826176166534423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,4,power_law_1.2,0.501855993270874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,4,1,power_law_1.2,6.342668914794922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,4,power_law_1.2,1.0240511894226074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,4,power_law_1.01,0.9037952423095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,4,power_law_1.01,9.632115173339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,4,1,balanced,0.06843733290831248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,4,1,balanced,0.08593599994977315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,4,power_law_1.2,1.106438446044922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,4,1,balanced,0.12459733088811238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,4,1,balanced,0.24898666143417358
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,4,power_law_1.01,1.1756095886230469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,4,1,balanced,0.44434134165445965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,4,power_law_1.01,0.5454527854919433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,4,power_law_1.2,1.1711359977722169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,4,power_law_1.01,0.5467008113861084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,4,power_law_1.2,0.09983999729156494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,4,power_law_1.2,1.337939167022705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,4,power_law_1.01,2.039232063293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,4,power_law_1.01,0.5804927825927735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,4,power_law_1.2,0.15756160020828247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,4,power_law_1.2,0.5567552089691162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,4,1,balanced,0.060565332571665444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,4,power_law_1.2,0.18247040510177612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,4,power_law_1.01,0.597331190109253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,4,power_law_1.2,0.5749120235443115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,4,power_law_1.2,1.5628607749938965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,4,power_law_1.2,0.24044160842895507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,4,power_law_1.01,9.611366271972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,4,power_law_1.01,0.6298111915588379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,4,power_law_1.2,0.5807360172271728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,4,power_law_1.2,0.3394047975540161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,4,1,balanced,0.8179199695587158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,4,power_law_1.01,0.6615295886993409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,4,power_law_1.2,0.6161791801452636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,4,power_law_1.2,1.8305791854858398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,4,power_law_1.2,0.42490878105163576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,4,power_law_1.01,0.7628032207489014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,4,power_law_1.2,0.5394688129425049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,4,1,balanced,0.050698667764663696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,4,power_law_1.01,0.8659584045410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,4,1,balanced,0.07626666625340779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,4,power_law_1.2,0.5080639839172363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,4,1,balanced,0.10081600149472554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,4,1,balanced,0.17966399590174356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,4,1,balanced,0.3067466616630554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,4,power_law_1.01,1.0829567909240723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,4,power_law_1.2,0.4904128074645996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,4,1,balanced,0.5749280055363973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,4,1,balanced,0.5764853159586588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,4,power_law_1.2,0.5069568157196045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,4,1,balanced,0.57860799630483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,4,power_law_1.01,1.348249626159668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,4,1,balanced,0.8225173155466715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,4,1,balanced,0.5809653202692667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,4,power_law_1.2,0.5498688220977783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,4,power_law_1.2,0.6532415866851806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,4,power_law_1.2,0.5375423908233643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,4,power_law_1.2,0.6710336208343506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,4,power_law_1.01,1.7438400268554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,4,power_law_1.2,0.5732927799224854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,4,power_law_1.2,2.4660032272338865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,4,power_law_1.2,0.7224127769470214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,4,power_law_1.2,0.5874879837036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,4,power_law_1.2,0.7487936019897461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,4,power_law_1.01,2.1330944061279298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,4,power_law_1.2,0.598419189453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,4,power_law_1.2,0.8294848442077637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,4,power_law_1.2,0.05146880149841308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,4,power_law_1.2,0.6259007930755616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,4,1,balanced,0.5806346734364828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,4,power_law_1.2,0.07500159740447998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,4,1,balanced,0.5864266554514567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,4,power_law_1.2,0.9691840171813965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,4,power_law_1.2,0.7006976127624511
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,4,power_law_1.2,0.08343679904937744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,4,power_law_1.01,3.042598342895508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,4,power_law_1.2,0.10534399747848511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,4,power_law_1.2,1.1283967971801758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,4,power_law_1.2,0.13568639755249023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,4,power_law_1.2,1.357695960998535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,4,power_law_1.2,0.15648640394210817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,4,power_law_1.2,3.1615999221801756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,4,1,balanced,0.8221440315246582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,4,power_law_1.01,3.944998550415039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,4,power_law_1.2,1.6999231338500977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,4,1,balanced,0.8277119795481364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,4,1,balanced,0.5909759998321533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,4,1,balanced,0.8291680018107096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,4,1,balanced,0.5940639972686768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,4,power_law_1.2,0.7898816108703614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,4,1,balanced,0.8306826750437418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,4,1,balanced,0.60371200243632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,4,1,balanced,0.8355466524759928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,4,power_law_1.2,2.3049600601196287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,4,power_law_1.2,0.9186047554016114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,4,1,balanced,0.8381546338399252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,4,1,balanced,0.8462613423665365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,4,power_law_1.2,0.21242880821228027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,4,1,balanced,0.9012853304545084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,4,power_law_1.01,5.624448013305664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,4,1,balanced,0.9167306423187256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,4,power_law_1.2,0.18565759658813477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,4,power_law_1.2,2.8618879318237305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,4,1,balanced,0.930560032526652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,4,power_law_1.2,0.18991999626159667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,4,1,balanced,0.9719733397165934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,4,power_law_1.2,0.2043776035308838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,4,1,balanced,1.014367977778117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,4,power_law_1.2,4.196627044677735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,4,1,balanced,1.1099733511606853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,4,power_law_1.2,4.095564651489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,4,1,balanced,1.2117493152618408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,4,power_law_1.2,1.1041152000427246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,4,1,balanced,1.691088040669759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,4,1,balanced,0.06807999809583028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,4,1,balanced,0.08840533097585042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,4,power_law_1.2,1.406118392944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,4,1,balanced,0.13473600149154663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,4,1,balanced,1.8989653587341309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,4,1,balanced,0.6206666628519694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,4,power_law_1.2,5.264793777465821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,4,1,balanced,0.630672017733256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,4,power_law_1.2,0.22288639545440675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,4,power_law_1.2,1.73504638671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,4,1,balanced,2.6619466145833335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,4,power_law_1.2,5.429433441162109
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,4,power_law_1.2,0.20414719581604004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,4,power_law_1.2,0.2149951934814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,4,power_law_1.2,0.230131196975708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,4,power_law_1.2,2.2822975158691405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,4,1,balanced,3.4793386459350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,4,power_law_1.2,0.2291264057159424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,4,power_law_1.01,7.326739501953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,4,power_law_1.2,0.23879680633544922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,4,power_law_1.2,0.2606656074523926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,4,power_law_1.2,0.2879807949066162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,4,1,balanced,0.651365319887797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,4,power_law_1.2,3.1326656341552734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,4,1,balanced,5.080117225646973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,4,1,balanced,0.7035306294759115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,4,power_law_1.2,8.527903747558593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,4,1,balanced,0.7548853556315104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,4,power_law_1.2,7.8395263671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,4,1,balanced,0.8677226702372233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,4,1,balanced,0.9873546759287516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,4,1,balanced,1.4598026275634766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,4,power_law_1.2,4.213343811035156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,4,1,balanced,7.007162729899089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,4,1,balanced,1.6960533459981282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,4,power_law_1.2,0.3008064031600952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,4,power_law_1.2,0.3731328010559082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,4,1,balanced,2.4105440775553384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,4,1,power_law_1.01,0.0670144021511078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,4,1,power_law_1.01,0.08376960158348083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,4,1,power_law_1.01,0.12183040380477905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,4,1,balanced,3.125077247619629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,4,power_law_1.2,5.808678436279297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,4,1,power_law_1.01,0.24442880153656005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,4,1,power_law_1.01,0.33815679550170896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,4,1,power_law_1.01,0.4653439998626709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,4,1,balanced,4.5709654490153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,4,power_law_1.2,10.299116516113282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,4,1,power_law_1.01,0.6206912040710449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,4,1,power_law_1.01,0.05092480182647705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,4,1,power_law_1.01,0.648902416229248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,4,1,power_law_1.01,0.0757311999797821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,4,1,power_law_1.01,0.6524543762207031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,4,1,power_law_1.01,0.09884160161018371
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,4,1,balanced,0.029605334003766377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,4,1,balanced,0.03379199902216593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,4,1,balanced,0.05619200070699056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,4,1,power_law_1.01,0.672166395187378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,4,1,balanced,0.08006399869918823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,4,1,balanced,0.12821333607037863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,4,1,balanced,0.2215786576271057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,4,1,power_law_1.01,0.6865471839904785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,4,1,power_law_1.01,0.7400320053100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,4,1,power_law_1.01,0.745849609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,4,power_law_1.2,0.41214718818664553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,4,1,power_law_1.01,0.8271103858947754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,4,1,power_law_1.01,0.1751680016517639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,4,power_law_1.2,7.926604461669922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,4,power_law_1.2,0.5349120140075684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,4,1,power_law_1.01,0.24822399616241456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,4,1,power_law_1.01,0.8478079795837402
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,4,1,balanced,0.2242293357849121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,4,power_law_1.2,0.5688767910003663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,4,1,balanced,0.23061867554982504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,4,1,power_law_1.01,0.31607680320739745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,4,1,balanced,0.23201066255569458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,4,1,power_law_1.01,0.8717823982238769
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,4,1,balanced,0.23462400833765665
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,4,1,balanced,0.23834667603174844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,4,1,power_law_1.01,0.4414400100708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,4,1,power_law_1.01,0.8911552429199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,4,1,power_law_1.01,0.4546495914459229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,4,1,balanced,6.0171254475911455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,4,1,power_law_1.01,0.04812160134315491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,4,1,power_law_1.01,0.9931776046752929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,4,1,power_law_1.01,0.4626304149627686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,4,1,balanced,0.24387733141581217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,4,1,power_law_1.01,0.0675711989402771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,4,1,balanced,0.4510186513264974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,4,1,power_law_1.01,0.477945613861084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,4,1,power_law_1.01,1.0553664207458495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,4,1,balanced,0.4516479969024658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,4,1,power_law_1.01,0.5044095993041993
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,4,power_law_1.2,0.7146687984466553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,4,1,power_law_1.01,0.550713586807251
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,4,1,balanced,0.24365333716074625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,4,1,balanced,0.24762133757273355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,4,1,balanced,0.258026659488678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,4,1,power_law_1.01,0.5727295875549316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,4,1,balanced,0.28148265679677326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,4,1,balanced,0.29016000032424927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,4,1,balanced,0.31092266241709393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,4,1,power_law_1.01,0.6395455837249756
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,4,1,balanced,0.31785066922505695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,4,1,balanced,0.3383786678314209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,4,1,power_law_1.01,0.5894847869873047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,4,1,balanced,0.3739360173543294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,4,1,power_law_1.01,0.0871295988559723
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,4,1,balanced,0.41018664836883545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,4,1,power_law_1.01,0.6104000091552735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,4,1,power_law_1.01,0.13192960023880004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,4,1,balanced,0.5070506731669108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,4,1,power_law_1.01,1.2240832328796387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,4,1,power_law_1.01,0.6496128082275391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,4,1,power_law_1.01,0.20833280086517333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,4,1,power_law_1.01,0.7355711936950684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,4,power_law_1.2,0.9017919540405274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,4,1,power_law_1.01,1.3863743782043456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,4,1,power_law_1.01,0.8196800231933594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,4,power_law_1.2,1.2461952209472655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,4,power_law_1.2,9.534111785888673
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,4,1,power_law_1.01,0.027820798754692077
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,4,1,power_law_1.01,0.03251200020313263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,4,1,power_law_1.2,0.06687359809875489
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,4,1,balanced,0.5342239936192831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,4,1,power_law_1.01,0.05260159969329834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,4,power_law_1.2,2.395552062988281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,4,1,balanced,0.999941349029541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,4,1,power_law_1.01,0.07738239765167236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,4,1,power_law_1.01,0.24739840030670165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,4,1,balanced,0.8017706871032715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,4,1,power_law_1.01,0.10485759973526002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,4,1,power_law_1.01,0.34119040966033937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,4,1,balanced,1.1249547004699707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,4,1,power_law_1.01,0.14118399620056152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,4,1,power_law_1.01,1.743244743347168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,4,1,power_law_1.01,0.17711999416351318
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,4,1,balanced,1.463919957478841
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,4,1,power_law_1.01,0.18951679468154908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,4,1,power_law_1.01,0.19328639507293702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,4,1,power_law_1.01,2.089676856994629
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,4,1,power_law_1.01,0.19791359901428224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,4,1,power_law_1.01,0.20455679893493653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,4,1,power_law_1.2,0.0491456001996994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,4,1,power_law_1.01,0.2132863998413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,4,1,power_law_1.2,0.07497599720954895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,4,1,power_law_1.2,0.05084159970283508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,4,1,power_law_1.01,2.87652473449707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,4,1,power_law_1.01,0.22385919094085693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,4,1,power_law_1.01,0.3574336051940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,4,1,power_law_1.01,0.24008960723876954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,4,1,power_law_1.01,0.365011191368103
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,4,1,power_law_1.01,0.26280961036682127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,4,1,power_law_1.2,0.08366079926490784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,4,1,power_law_1.01,0.3732736110687256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,4,1,power_law_1.01,0.27729918956756594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,4,1,power_law_1.2,0.12358399629592895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,4,1,power_law_1.01,0.3792448043823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,4,1,power_law_1.01,0.3215487957000732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,4,1,power_law_1.01,3.5598655700683595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,4,1,power_law_1.2,0.23081600666046143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,4,1,power_law_1.01,0.41910400390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,4,1,power_law_1.01,0.3121151924133301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,4,1,power_law_1.2,0.3174720048904419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,4,1,power_law_1.01,0.42646398544311526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,4,1,power_law_1.01,0.36214399337768555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,4,1,power_law_1.01,0.9923135757446289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,4,1,power_law_1.2,0.41164798736572267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,4,1,power_law_1.01,0.37420799732208254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,4,1,power_law_1.01,1.1727744102478028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,4,1,balanced,0.4524213473002116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,4,1,power_law_1.01,5.34901123046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,4,1,power_law_1.01,1.5273280143737793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,4,1,balanced,0.4585973421732585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,4,1,balanced,0.4591093460718791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,4,1,balanced,0.4650346835454305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,4,1,balanced,0.4689120054244995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,4,1,power_law_1.2,0.0982912003993988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,4,1,power_law_1.01,1.8946239471435546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,4,1,power_law_1.01,0.4463679790496826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,4,1,power_law_1.2,0.1786687970161438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,4,1,power_law_1.2,0.6150207996368409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,4,1,power_law_1.2,0.22239360809326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,4,1,power_law_1.2,0.2966912031173706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,4,1,power_law_1.2,0.42266240119934084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,4,1,power_law_1.01,6.984454345703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,4,1,power_law_1.2,0.44219517707824707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,4,1,balanced,0.4742240111033122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,4,1,balanced,0.4801599979400635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,4,1,balanced,0.4904266595840454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,4,1,power_law_1.01,0.4873663902282715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,4,1,balanced,0.4960586627324422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,4,1,balanced,0.5890026489893595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,4,1,power_law_1.01,2.602400016784668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,4,1,balanced,0.6179466644922892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,4,1,balanced,0.7144213517506918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,4,1,balanced,0.7791039943695068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,4,1,balanced,0.8851199944814047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,4,1,power_law_1.2,0.46822400093078614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,4,1,power_law_1.01,3.3351295471191404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,4,1,balanced,1.3355573018391926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,4,1,power_law_1.01,0.4460031986236572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,4,1,power_law_1.2,0.4814591884613037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,4,1,balanced,1.4964799880981445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,4,1,power_law_1.01,0.47757439613342284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,4,1,power_law_1.2,0.5134399890899658
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,4,1,power_law_1.01,0.5806464195251465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,4,1,balanced,2.118058681488037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,4,1,power_law_1.01,0.7220352172851563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,4,1,power_law_1.01,4.763692855834961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,4,1,balanced,2.7428372701009116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,4,1,power_law_1.01,0.8968640327453613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,4,1,power_law_1.2,0.06784639954566955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,4,1,power_law_1.2,0.028966400027275085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,4,1,power_law_1.2,0.08792319893836975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,4,1,power_law_1.01,0.5177472114562989
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,4,1,power_law_1.2,0.03436160087585449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,4,1,power_law_1.2,0.6517568111419678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,4,1,power_law_1.2,0.12373119592666626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,4,1,power_law_1.01,1.227468776702881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,4,1,balanced,4.057221412658691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,4,1,power_law_1.01,0.6261951923370361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,4,1,power_law_1.2,0.6448895931243896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,4,1,power_law_1.01,0.7720640182495118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,4,1,power_law_1.2,0.6640448093414306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,4,1,power_law_1.01,1.7587711334228515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,4,1,power_law_1.01,6.208243179321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,4,balanced,0.027056001126766205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,4,balanced,0.030159999926884968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,4,balanced,0.029258665939172108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,4,balanced,0.031557333966096245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,4,balanced,0.039749334255854286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,4,1,power_law_1.01,0.7651135921478271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,4,balanced,0.05565866827964783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,4,balanced,0.07972266773382823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,4,balanced,0.0806933343410492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,4,balanced,0.09438400467236836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,4,balanced,0.09495466947555542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,4,balanced,0.09505066275596619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,4,balanced,0.09899200002352397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,4,balanced,0.10014933347702026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,4,balanced,0.10282133022944133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,4,balanced,0.10520533720652263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,4,balanced,0.10834667086601257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,4,balanced,0.11353066563606262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,4,balanced,0.12103999654452006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,4,balanced,0.12944533427556357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,4,balanced,0.1481013298034668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,4,balanced,0.1697333256403605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,4,balanced,0.24499199787775675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,4,balanced,0.2899786631266276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,4,balanced,0.03772799919048945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,4,balanced,0.05237866441408793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,4,balanced,0.0747626672188441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,4,balanced,0.0746666689713796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,4,balanced,0.0888853371143341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,4,balanced,0.08972799777984619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,4,1,power_law_1.01,0.9204480171203613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,4,balanced,0.09082133571306865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,4,balanced,0.09172266721725464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,4,balanced,0.09646933277448018
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,4,1,power_law_1.2,0.05305600166320801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,4,1,power_law_1.2,0.5553343772888184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,4,1,power_law_1.01,1.0751359939575196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,4,1,power_law_1.2,0.07318400144577027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,4,1,power_law_1.2,0.5884416103363037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,4,1,power_law_1.2,0.0966592013835907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,4,1,power_law_1.2,0.6572415828704834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,4,1,power_law_1.01,1.3867775917053222
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,4,1,power_law_1.2,0.1309056043624878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,4,1,power_law_1.2,0.6828288078308106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,4,balanced,0.40223999818166095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,4,1,balanced,5.273749351501465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,4,1,power_law_1.2,0.6046912193298339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,4,balanced,0.5071946779886881
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,4,1,power_law_1.2,0.1777151942253113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,4,1,power_law_1.2,0.7402944087982177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,4,balanced,0.037834666669368744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,4,balanced,0.044026667873064675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,4,balanced,0.730778694152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,4,1,power_law_1.2,0.6175936222076416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,4,1,power_law_1.01,1.702284812927246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,4,1,power_law_1.2,0.187116801738739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,4,balanced,0.09930666287740071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,4,1,power_law_1.2,0.7803391933441162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,4,balanced,0.1011786659558614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,4,balanced,0.9514559904734293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,4,balanced,0.1072160005569458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,4,1,power_law_1.2,0.19452799558639527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,4,1,power_law_1.2,0.6602176189422607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,4,1,power_law_1.2,0.851961612701416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,4,balanced,1.3986239433288574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,4,1,power_law_1.2,0.19676799774169923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,4,1,power_law_1.2,0.7590784072875977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,4,1,power_law_1.01,2.34192008972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,4,1,power_law_1.2,0.20524160861968993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,4,1,power_law_1.2,0.8659520149230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,4,balanced,1.8424852689107258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,4,1,power_law_1.2,0.21568000316619873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,4,1,power_law_1.2,0.8514304161071777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,4,1,power_law_1.2,0.8887167930603027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,4,1,power_law_1.2,1.0217151641845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,4,balanced,0.04566933214664459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,4,balanced,0.04990933338801066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,4,balanced,0.08248533308506012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,4,balanced,0.0817386656999588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,4,balanced,0.0827893316745758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,4,balanced,0.11296000083287557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,4,balanced,0.08515733480453491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,4,balanced,0.11923199892044067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,4,1,power_law_1.2,1.2006464004516602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,4,balanced,0.08584533135096233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,4,balanced,0.1288266678651174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,4,balanced,0.08906132976214091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,4,balanced,0.15058133006095886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,4,balanced,0.09014399846394856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,4,balanced,0.17809067169825235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,4,balanced,0.09327999750773112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,4,balanced,0.09653866291046143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,4,balanced,0.10099200407663982
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,4,balanced,0.0234400009115537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,4,balanced,0.10750933488210042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,4,balanced,0.02609066665172577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,4,balanced,0.11120532949765523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,4,balanced,0.027109332382678986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,4,balanced,0.12226667006810506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,4,balanced,0.14199466506640115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,4,balanced,0.16412267088890076
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,4,1,power_law_1.2,0.22449920177459717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,4,balanced,0.2307200034459432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,4,balanced,0.2754720052083333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,4,1,power_law_1.2,0.24689919948577882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,4,balanced,0.38630934556325275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,4,1,power_law_1.2,0.9572671890258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,4,1,power_law_1.01,2.975916862487793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,4,1,power_law_1.2,0.26657280921936033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,4,balanced,0.49373332659403485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,4,balanced,0.707856019337972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,4,1,power_law_1.2,1.0330495834350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,4,balanced,0.9245920181274414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,4,balanced,0.2467306653658549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,4,balanced,0.2959413329760234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,4,1,power_law_1.2,1.093344020843506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,4,1,power_law_1.2,1.5617919921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,4,balanced,0.029525332152843475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,4,balanced,0.03469866762558619
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,4,balanced,0.03530666728814443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,4,balanced,0.0376800000667572
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,4,balanced,0.03910933434963226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,4,1,power_law_1.01,4.229836654663086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,4,balanced,0.03976533313592275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,4,1,power_law_1.2,1.2515583992004395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,4,1,power_law_1.2,1.9212928771972657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,4,1,power_law_1.2,0.28330240249633787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,4,1,power_law_1.2,1.4246399879455567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,4,1,power_law_1.2,0.33056640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,4,1,power_law_1.2,0.19741439819335938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,4,1,power_law_1.2,2.6412736892700197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,4,balanced,1.3502826690673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,4,balanced,0.4160906473795573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,4,1,power_law_1.2,1.7725183486938476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,4,1,power_law_1.2,0.23320960998535156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,4,balanced,0.5320266485214233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,4,1,power_law_1.01,5.492972946166992
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,4,balanced,0.03975466638803482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,4,balanced,0.040218666195869446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,4,balanced,0.042863999803860985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,4,power_law_1.01,0.03450239896774292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,4,balanced,0.043882668018341064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,4,1,power_law_1.2,2.134796714782715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,4,balanced,0.042117332418759666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,4,power_law_1.01,0.042368000745773314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,4,1,power_law_1.2,3.367571258544922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,4,power_law_1.01,0.05306879878044128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,4,power_law_1.01,0.05764480233192444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,4,power_law_1.01,0.067084801197052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,4,1,power_law_1.2,0.3230783939361572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,4,power_law_1.01,0.07403519749641418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,4,1,power_law_1.2,2.9062271118164062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,4,power_law_1.01,0.09034879803657532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,4,1,power_law_1.2,0.3761023998260498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,4,power_law_1.01,0.09182080030441284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,4,balanced,1.782474676767985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,4,power_law_1.01,0.09471359848976135
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,4,1,power_law_1.2,0.38771839141845704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,4,power_law_1.01,0.032441601157188416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,4,1,power_law_1.2,4.812748718261719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,4,1,power_law_1.2,0.45720319747924804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,4,power_law_1.01,0.040300801396369934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,4,balanced,0.05333333214124044
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,4,balanced,0.05003733436266581
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,4,balanced,0.04828799764315287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,4,balanced,0.07872533301512401
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,4,1,power_law_1.2,0.49088001251220703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,4,balanced,0.06681600213050842
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,4,balanced,0.07458666463692983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,4,balanced,0.08136533200740814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,4,balanced,0.10947199662526448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,4,1,power_law_1.2,0.5878911972045898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,4,balanced,0.12341333429018657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,4,balanced,0.16672533750534058
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,4,balanced,0.20993600289026895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,4,1,power_law_1.2,0.7485695838928222
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,4,balanced,0.2961440086364746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,4,balanced,0.38789331912994385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,4,power_law_1.01,0.09396479725837707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,4,1,power_law_1.2,0.9067520141601563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,4,power_law_1.01,0.09761919975280761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,4,1,power_law_1.2,6.260908889770508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,4,power_law_1.01,0.03671680092811584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,4,power_law_1.01,0.10058879852294922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,4,power_law_1.01,0.045465600490570066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,4,power_law_1.01,0.046086400747299194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,4,power_law_1.01,0.052153599262237546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,4,1,power_law_1.2,1.2397631645202636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,4,power_law_1.01,0.05457280278205871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,4,power_law_1.01,0.06718720197677612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,4,power_law_1.01,0.08401920199394226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,4,power_law_1.01,0.08675839900970458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,4,1,power_law_1.2,1.8125247955322266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,4,power_law_1.01,0.08841599822044373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,4,power_law_1.01,0.08907520174980163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,4,power_law_1.2,0.032979199290275575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,4,power_law_1.01,0.09285119771957398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,4,power_law_1.2,0.04439679980278015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,4,power_law_1.01,0.09749119877815246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,4,1,power_law_1.2,3.73306884765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,4,power_law_1.01,0.1012992024421692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,4,power_law_1.01,0.11294080018997192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,4,power_law_1.01,0.11695359945297241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,4,power_law_1.01,0.12616319656372071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,4,balanced,0.7671253681182861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,4,power_law_1.01,0.14305919408798218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,4,balanced,1.000111977259318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,4,power_law_1.01,0.16851840019226075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,4,power_law_1.01,0.1982975959777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,4,power_law_1.01,0.02333440035581589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,4,1,power_law_1.2,5.241209411621094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,4,power_law_1.01,0.2637631893157959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,4,power_law_1.01,0.10279040336608887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,4,power_law_1.01,0.025798401236534117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,4,power_law_1.01,0.11083519458770752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,4,power_law_1.01,0.32780799865722654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,4,power_law_1.01,0.11660159826278686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,4,power_law_1.01,0.4623104095458984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,4,power_law_1.01,0.12645119428634644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,4,power_law_1.01,0.13812479972839356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,4,power_law_1.01,0.5779391765594483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,4,power_law_1.01,0.17635200023651124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,4,power_law_1.01,0.20007040500640869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,4,power_law_1.01,0.8481216430664062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,4,power_law_1.01,0.2596544027328491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,4,balanced,1.4673333168029785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,4,power_law_1.01,0.31636478900909426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,4,power_law_1.01,1.0709247589111328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,4,1,power_law_1.2,6.9928642272949215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,4,power_law_1.01,0.44878082275390624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,4,balanced,1.9363093376159668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,4,power_law_1.2,0.034220799803733826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,4,power_law_1.01,0.5714176177978516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,4,power_law_1.01,1.6375423431396485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,4,power_law_1.2,0.04279040098190308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,4,power_law_1.01,0.7968575954437256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,4,power_law_1.01,1.0294591903686523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,4,power_law_1.2,0.04935039877891541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,4,1,power_law_1.2,0.3381311893463135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,4,power_law_1.01,0.050329601764678954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,4,power_law_1.2,0.0580735981464386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,4,power_law_1.01,1.5806079864501954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,4,power_law_1.01,0.055692797899246214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,4,power_law_1.01,0.0667136013507843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,4,power_law_1.2,0.03777920007705689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,4,power_law_1.01,0.07795199751853943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,4,power_law_1.2,0.04333440065383911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,4,power_law_1.01,2.0897920608520506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,4,power_law_1.01,0.07954559922218322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,4,power_law_1.2,0.049728000164031984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,4,power_law_1.01,0.08268160223960877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,4,power_law_1.2,0.05480319857597351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,4,power_law_1.01,2.1394880294799803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,4,power_law_1.01,0.08256000280380249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,4,1,balanced,0.021162666380405426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,4,1,balanced,0.023141334454218548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,4,power_law_1.2,0.06704639792442321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,4,power_law_1.01,0.08403199911117554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,4,power_law_1.2,0.07727360129356384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,4,power_law_1.01,0.08750720024108886
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,4,power_law_1.01,0.026700800657272337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,4,power_law_1.2,0.08101119995117187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,4,power_law_1.01,0.09006080031394958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,4,power_law_1.2,0.05564799904823303
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,4,power_law_1.01,0.028505599498748778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,4,power_law_1.2,0.08208640217781067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,4,power_law_1.01,0.09201279878616334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,4,power_law_1.01,0.029708799719810487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,4,power_law_1.01,0.09947519898414611
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,4,power_law_1.01,0.03356159925460815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,4,power_law_1.2,0.0469760000705719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,4,power_law_1.01,0.10645120143890381
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,4,power_law_1.2,0.02421119958162308
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,4,power_law_1.01,0.03691520094871521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,4,power_law_1.2,0.05081599950790405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,4,power_law_1.01,0.11391359567642212
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,4,power_law_1.2,0.026118400692939758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,4,power_law_1.2,0.05188480019569397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,4,power_law_1.01,0.12700159549713136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,4,power_law_1.2,0.027609598636627198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,4,power_law_1.2,0.06372479796409607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,4,power_law_1.01,0.1560639977455139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,4,power_law_1.2,0.08651520013809204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,4,power_law_1.2,0.08595200181007386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,4,power_law_1.2,0.08857600092887878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,4,power_law_1.2,0.08391039967536926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,4,power_law_1.2,0.09048320055007934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,4,power_law_1.2,0.08574720025062561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,4,power_law_1.2,0.0933184027671814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,4,power_law_1.2,0.08710399866104127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,4,power_law_1.2,0.10037120580673217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,4,power_law_1.01,0.039852800965309146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,4,power_law_1.2,0.09111679792404175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,4,power_law_1.2,0.1039423942565918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,4,power_law_1.01,0.04062080085277557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,4,power_law_1.2,0.09401599764823913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,4,power_law_1.2,0.11374080181121826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,4,1,balanced,0.033386667569478355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,4,1,balanced,0.05393599967161814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,4,power_law_1.2,0.029100799560546876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,4,power_law_1.2,0.10012799501419067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,4,power_law_1.2,0.11777280569076538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,4,power_law_1.2,0.10794880390167236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,4,power_law_1.2,0.12832000255584716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,4,power_law_1.01,0.18622080087661744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,4,power_law_1.2,0.11521279811859131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,4,power_law_1.2,0.1449087977409363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,4,power_law_1.2,0.07222399711608887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,4,power_law_1.01,0.24407680034637452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,4,power_law_1.2,0.17575039863586425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,4,power_law_1.2,0.09187840223312378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,4,power_law_1.01,0.29719679355621337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,4,power_law_1.2,0.09438719749450683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,4,power_law_1.2,0.09449599981307984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,4,power_law_1.2,0.09729920029640197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,4,power_law_1.2,0.0992959976196289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,4,power_law_1.01,0.04063999950885773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,4,1,balanced,0.07845866680145264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,4,power_law_1.2,0.10267519950866699
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,4,power_law_1.01,0.042080000042915344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,4,power_law_1.2,0.02998400032520294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,4,power_law_1.2,0.10721280574798583
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,4,power_law_1.01,0.04336639940738678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,4,power_law_1.2,0.1143231987953186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,4,power_law_1.01,0.04423680007457733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,4,power_law_1.2,0.1289471983909607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,4,power_law_1.2,0.12446080446243286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,4,power_law_1.01,0.04749439954757691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,4,power_law_1.2,0.15943679809570313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,4,power_law_1.2,0.2044991970062256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,4,power_law_1.01,0.04871039986610413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,4,power_law_1.2,0.1894976019859314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,4,power_law_1.2,0.2698240041732788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,4,power_law_1.01,0.42885122299194334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,4,power_law_1.01,0.04913919866085052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,4,power_law_1.2,0.2524928092956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,4,power_law_1.2,0.34471681118011477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,4,power_law_1.01,0.05248000025749207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,4,power_law_1.01,0.5535295963287353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,4,power_law_1.2,0.30899200439453123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,4,power_law_1.01,0.07427840232849121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,4,1,balanced,0.08044800162315369
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,4,power_law_1.01,0.07173759937286377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,4,1,balanced,0.08183999856313069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,4,power_law_1.2,0.4282112121582031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,4,1,balanced,0.0803466687599818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,4,1,balanced,0.08075733482837677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,4,power_law_1.2,0.034720000624656674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,4,power_law_1.2,0.5622848033905029
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,4,power_law_1.2,0.03854719996452331
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,4,power_law_1.2,0.04042240083217621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,4,power_law_1.2,0.12910720109939575
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,4,power_law_1.2,0.040454399585723874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,4,power_law_1.2,0.15148799419403075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,4,power_law_1.2,0.042419201135635375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,4,power_law_1.2,0.17916159629821776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,4,power_law_1.2,0.04337919950485229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,4,power_law_1.2,0.20683519840240477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,4,1,power_law_1.2,0.35508480072021487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,4,power_law_1.2,0.2730943918228149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,4,power_law_1.01,0.7569280147552491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,4,1,power_law_1.2,0.3678015947341919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,4,power_law_1.01,0.08181120157241821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,4,power_law_1.2,0.3465536117553711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,4,1,balanced,0.08176533381144206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,4,1,balanced,0.08283733328183492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,4,1,balanced,0.09562666217486064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,4,1,power_law_1.2,0.3726720094680786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,4,1,balanced,0.09739733735720317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,4,power_law_1.01,1.0156991958618165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,4,1,balanced,0.10056533416112264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,4,1,balanced,0.10364266236623128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,4,1,balanced,0.10719466209411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,4,1,balanced,0.11315733194351196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,4,1,balanced,0.12575999895731607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,4,1,balanced,0.13874133427937826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,4,1,balanced,0.1908479928970337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,4,power_law_1.01,1.5164416313171387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,4,1,balanced,0.21818133195241293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,4,1,balanced,0.2908746600151062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,4,power_law_1.2,0.04382080137729645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,4,power_law_1.2,0.8061951637268067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,4,power_law_1.2,0.0460671991109848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,4,power_law_1.2,0.4635776042938232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,4,power_law_1.01,1.9663423538208007
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,4,power_law_1.2,0.04745599925518036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,4,1,balanced,0.0230880007147789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,4,1,balanced,0.025392000873883564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,4,power_law_1.2,0.04960640072822571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,4,1,balanced,0.029578665892283123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,4,power_law_1.2,0.48633599281311035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,4,power_law_1.2,0.050758397579193114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,4,1,power_law_1.2,0.39151360988616946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,4,power_law_1.2,0.05318400263786316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,4,power_law_1.2,0.5730112075805665
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,4,power_law_1.2,0.07294719815254211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,4,1,power_law_1.2,0.4188352108001709
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,4,power_law_1.2,0.07571840286254883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,4,1,power_law_1.2,0.42862720489501954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,4,1,balanced,0.35997335116068524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,4,1,balanced,0.5105013449986776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,4,1,power_law_1.2,0.4512063980102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,4,1,balanced,0.6617386738459269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,4,1,power_law_1.2,0.5069695949554444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,4,power_law_1.01,0.0945792019367218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,4,1,balanced,0.9678400357564291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,4,1,balanced,0.05003733436266581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,4,1,balanced,0.07734933495521545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,4,power_law_1.01,0.12051199674606324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,4,1,balanced,0.07628799974918365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,4,1,balanced,1.2717920144399006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,4,power_law_1.01,0.1396607995033264
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,4,power_law_1.01,0.195251202583313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,4,power_law_1.01,0.23741440773010253
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,4,power_law_1.2,0.08659200072288513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,4,power_law_1.2,0.0981760025024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,4,power_law_1.2,0.8265855789184571
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,4,1,balanced,0.025589334468046825
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,4,1,balanced,0.025279998779296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,4,1,power_law_1.2,0.5187712192535401
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,4,1,balanced,0.027263998985290527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,4,1,balanced,0.07874666651089986
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,4,1,balanced,0.03128000100453695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,4,1,balanced,0.07868266602357228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,4,1,balanced,0.07859733204046886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,4,1,balanced,0.07910933097203572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,4,1,balanced,0.07916800181070964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,4,1,balanced,0.09197866916656494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,4,1,balanced,0.09444800019264221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,4,1,balanced,0.09550399581591289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,4,1,balanced,0.10047466556231181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,4,1,balanced,0.10537599523862202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,4,power_law_1.2,0.6228415966033936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,4,power_law_1.01,0.3605439901351929
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,4,power_law_1.01,0.48049278259277345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,4,1,power_law_1.01,0.019193600118160247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,4,1,power_law_1.01,0.027827200293540955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,4,power_law_1.2,1.053708839416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,4,1,power_law_1.01,0.03607679903507233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,4,1,power_law_1.2,0.6686975955963135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,4,1,power_law_1.01,0.04639999866485596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,4,power_law_1.2,0.12224639654159546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,4,1,power_law_1.01,0.0532800018787384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,4,1,power_law_1.2,0.7957376003265381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,4,1,power_law_1.01,0.07088639736175537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,4,1,balanced,0.11210133632024129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,4,1,balanced,0.1291146675745646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,4,1,balanced,0.14422399799029031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,4,1,balanced,0.1909066637357076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,4,1,balanced,0.22214933236440024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,4,1,balanced,0.30190932750701904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,4,1,balanced,0.3789386749267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,4,1,balanced,0.03905066599448522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,4,1,balanced,0.5411146481831869
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,4,power_law_1.2,0.1478335976600647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,4,1,balanced,0.6968426704406738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,4,power_law_1.2,1.5224448204040528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,4,1,balanced,1.016208012898763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,4,1,power_law_1.01,0.07400959730148315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,4,1,balanced,1.3401494026184082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,4,1,power_law_1.01,0.07493759989738465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,4,1,power_law_1.01,0.021887999773025513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,4,1,power_law_1.01,0.07856000065803528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,4,1,power_law_1.01,0.029286399483680725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,4,1,power_law_1.01,0.08092799782752991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,4,1,power_law_1.01,0.03346560001373291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,4,1,balanced,0.04126933217048645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,4,1,power_law_1.01,0.04322560131549835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,4,1,balanced,0.04159999887148539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,4,1,balanced,0.04223466912905375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,4,1,balanced,0.04398400088151296
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,4,1,balanced,0.04523199796676636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,4,1,balanced,0.04619733492533366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,4,1,balanced,0.05771199862162272
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,4,1,balanced,0.057989334066708885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,4,1,balanced,0.061994666854540505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,4,1,balanced,0.06004266440868378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,4,1,balanced,0.06062399844328562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,4,1,balanced,0.06482666730880737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,4,1,power_law_1.2,0.787334394454956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,4,1,power_law_1.2,0.94967041015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,4,1,power_law_1.01,0.0862335979938507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,4,1,power_law_1.01,0.09425920248031616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,4,1,power_law_1.01,0.09624320268630981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,4,power_law_1.2,0.8297151565551758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,4,1,power_law_1.2,1.106227207183838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,4,1,power_law_1.01,0.09978240132331848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,4,1,power_law_1.01,0.10311679840087891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,4,1,power_law_1.01,0.10726399421691894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,4,1,power_law_1.2,1.4118592262268066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,4,1,power_law_1.01,0.11592320203781128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,4,1,balanced,0.0708000014225642
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,4,1,balanced,0.07646400233109792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,4,1,power_law_1.01,0.14125440120697022
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,4,1,balanced,0.09666132926940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,4,1,balanced,0.10508267084757487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,4,1,power_law_1.01,0.16427520513534546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,4,1,balanced,0.16148799657821655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,4,1,balanced,0.1611199975013733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,4,1,power_law_1.2,1.7251199722290038
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,4,1,balanced,0.22553600867589316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,4,1,balanced,0.28195200363794964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,4,1,balanced,0.3903306722640991
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,4,1,balanced,0.5091946522394816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,4,1,power_law_1.2,2.36496639251709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,4,1,power_law_1.01,0.1962048053741455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,4,1,power_law_1.2,2.9929792404174806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,4,1,power_law_1.01,0.051609599590301515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,4,1,power_law_1.01,0.023552000522613525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,4,1,power_law_1.01,0.06556159853935242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,4,1,power_law_1.01,0.02426239997148514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,4,1,power_law_1.01,0.06696959733963012
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,4,1,power_law_1.01,0.02600319981575012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,4,1,power_law_1.01,0.06780160069465638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,4,1,power_law_1.01,0.07517439723014832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,4,1,power_law_1.01,0.07774720191955567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,4,1,power_law_1.01,0.0825984001159668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,4,1,power_law_1.01,0.08869760036468506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,4,1,power_law_1.01,0.23539841175079346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,4,1,power_law_1.01,0.3094655990600586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,4,1,power_law_1.01,0.38801279067993166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,4,power_law_1.2,1.0817279815673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,4,1,power_law_1.01,0.5410367965698242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,4,1,power_law_1.01,0.02924799919128418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,4,power_law_1.2,2.008889579772949
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,4,1,power_law_1.01,0.032467201352119446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,4,1,power_law_1.01,0.6934336185455322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,4,1,power_law_1.2,0.019916799664497376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,4,1,power_law_1.2,4.247846221923828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,4,1,power_law_1.01,0.03694080114364624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,4,power_law_1.2,1.1659071922302247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,4,1,power_law_1.2,0.029120001196861266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,4,1,power_law_1.01,0.040217599272727965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,4,1,power_law_1.01,1.005497646331787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,4,1,power_law_1.2,0.03532159924507141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,4,1,power_law_1.01,0.09232640266418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,4,1,power_law_1.01,0.04286719858646393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,4,1,power_law_1.2,0.04488320052623749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,4,1,power_law_1.01,0.09619839787483216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,4,1,power_law_1.2,0.05100160241127014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,4,1,power_law_1.01,0.1028223991394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,4,1,power_law_1.01,1.3189184188842773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,4,1,power_law_1.2,0.07179520130157471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,4,1,power_law_1.01,0.11127040386199952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,4,1,power_law_1.2,0.022886399924755097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,4,1,power_law_1.2,0.07371519804000855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,4,1,power_law_1.01,0.12046079635620117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,4,1,power_law_1.2,0.025331199169158936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,4,1,power_law_1.01,0.14200960397720336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,4,1,power_law_1.2,5.49870719909668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,4,1,power_law_1.01,0.045484799146652224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,4,1,power_law_1.01,0.04579200148582459
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,4,1,power_law_1.01,0.04870400130748749
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,4,1,power_law_1.01,0.05231999754905701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,4,power_law_1.2,0.20273919105529786
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,4,1,power_law_1.01,0.05508480072021484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,4,1,power_law_1.2,0.035104000568389894
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,4,1,power_law_1.01,0.0606656014919281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,4,1,power_law_1.2,0.023948800563812257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,4,1,power_law_1.01,0.15939199924468994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,4,1,power_law_1.01,0.05539839863777161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,4,1,power_law_1.01,0.1991039991378784
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,4,1,power_law_1.01,0.05945600271224975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,4,1,power_law_1.01,0.24078080654144288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,4,1,power_law_1.01,0.06487039923667907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,4,1,power_law_1.01,0.3195456027984619
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,4,1,power_law_1.01,0.06943359971046448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,4,1,power_law_1.01,0.07738879919052125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,4,1,power_law_1.01,0.09379199743270875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,4,1,power_law_1.2,0.07656959891319275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,4,1,power_law_1.01,0.1106943964958191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,4,1,power_law_1.2,0.04199039936065674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,4,1,power_law_1.2,0.08042240142822266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,4,1,power_law_1.2,0.04663040041923523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,4,1,power_law_1.2,0.08161280155181885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,4,1,power_law_1.2,0.06672000288963317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,4,1,power_law_1.2,0.08910719752311706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,4,1,power_law_1.2,0.06744319796562195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,4,1,power_law_1.2,0.09365760087966919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,4,1,power_law_1.2,0.09617919921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,4,1,power_law_1.01,0.4020671844482422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,4,1,power_law_1.2,0.09939839839935302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,4,1,power_law_1.2,0.10295039415359497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,4,1,power_law_1.01,0.5649472236633301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,4,1,power_law_1.2,0.10837759971618652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,4,1,power_law_1.01,0.1430400013923645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,4,1,power_law_1.2,0.11711360216140747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,4,1,power_law_1.01,0.7276480197906494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,4,1,power_law_1.2,0.02492160052061081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,4,1,power_law_1.2,0.1502784013748169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,4,1,power_law_1.2,0.16577279567718506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,4,1,power_law_1.01,1.0581631660461426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,4,1,power_law_1.2,0.19774080514907838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,4,1,power_law_1.2,0.07272959947586059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,4,1,power_law_1.2,0.2402496099472046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,4,1,power_law_1.01,1.3903231620788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,4,1,power_law_1.2,0.31369600296020506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,4,balanced,0.044346665342648826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,4,balanced,0.0680320014556249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,4,balanced,0.0920960009098053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,4,balanced,0.1420960028966268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,4,balanced,0.24307199319203696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,4,1,power_law_1.2,0.3892927885055542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,4,balanced,0.2427519957224528
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,4,1,power_law_1.01,0.17221120595932007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,4,balanced,0.2749119997024536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,4,balanced,0.2762879927953084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,4,1,power_law_1.2,0.026214399933815004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,4,1,power_law_1.2,0.02871679961681366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,4,1,power_law_1.2,0.07772160172462464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,4,1,power_law_1.2,0.03271040022373199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,4,1,power_law_1.2,0.0795199990272522
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,4,1,power_law_1.2,0.037350401282310486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,4,1,power_law_1.2,0.08372480273246766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,4,1,power_law_1.2,0.0404992014169693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,4,1,power_law_1.2,0.08836479783058167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,4,1,power_law_1.2,0.04273279905319214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,4,1,power_law_1.2,0.09400960206985473
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,4,1,power_law_1.2,0.044409599900245664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,4,1,power_law_1.2,0.09676160216331482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,4,1,power_law_1.2,0.04545280039310455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,4,1,power_law_1.2,0.10192639827728271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,4,balanced,0.2797386646270752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,4,1,power_law_1.2,0.048607999086380006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,4,balanced,0.28296534220377606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,4,1,power_law_1.2,0.11134719848632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,4,balanced,0.2867306669553121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,4,1,power_law_1.2,0.051744002103805545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,4,power_law_1.2,0.25845119953155515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,4,balanced,0.2959360082944234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,4,balanced,0.29981333017349243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,4,balanced,0.3097440004348755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,4,balanced,0.32205865780512494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,4,balanced,0.3339093526204427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,4,balanced,0.3564266761144002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,4,balanced,0.4002186854680379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,4,balanced,0.44787200291951496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,4,1,power_law_1.2,0.543507194519043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,4,balanced,0.6126346588134766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,4,1,power_law_1.01,0.23199360370635985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,4,balanced,0.7165119647979736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,4,1,power_law_1.01,0.28928000926971437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,4,balanced,1.022485335667928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,4,1,power_law_1.2,0.12008320093154908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,4,1,power_law_1.2,0.05494400262832642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,4,1,power_law_1.2,0.14371199607849122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,4,1,power_law_1.01,0.3994623899459839
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,4,1,power_law_1.2,0.060787200927734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,4,1,power_law_1.2,0.16076799631118774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,4,1,power_law_1.2,0.056595200300216676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,4,1,power_law_1.01,0.5198912143707275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,4,1,power_law_1.2,0.2020927906036377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,4,balanced,0.04167466859022776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,4,1,power_law_1.2,0.059008002281188965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,4,balanced,0.049653331438700356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,4,balanced,0.07745600243409474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,4,balanced,0.11554666360219319
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,4,1,power_law_1.2,0.06477439999580384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,4,balanced,0.1750826636950175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,4,balanced,0.1755519906679789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,4,1,power_law_1.2,0.07002239823341369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,4,balanced,0.21012800931930542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,4,1,power_law_1.2,0.24175360202789306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,4,balanced,0.2127359906832377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,4,1,power_law_1.2,0.07787520289421082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,4,balanced,0.21620800097783408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,4,balanced,0.21971199909845987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,4,balanced,0.22380266586939493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,4,balanced,0.23027199506759644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,4,balanced,0.23990400632222494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,4,balanced,0.25221866369247437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,4,balanced,0.2634399930636088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,4,balanced,1.310634692509969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,4,balanced,0.27618134021759033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,4,balanced,0.3028266628583272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,4,balanced,0.35382401943206787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,4,1,power_law_1.2,0.6972608089447021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,4,balanced,0.4101173480351766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,4,balanced,1.9183252652486165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,4,balanced,0.6019839843114217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,4,balanced,0.7143413225809733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,4,1,power_law_1.2,0.32235519886016845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,4,balanced,2.5090559323628745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,4,balanced,1.0205012957255046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,4,1,power_law_1.2,0.0938431978225708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,4,balanced,1.324618657430013
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,4,1,power_law_1.2,0.10922240018844605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,4,balanced,1.930448055267334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,4,1,power_law_1.2,0.14347519874572753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,4,balanced,3.981072107950846
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,4,1,power_law_1.2,0.1738752007484436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,4,balanced,2.5366133054097495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,4,balanced,5.569461186726888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,4,balanced,3.749312082926432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,4,balanced,0.05596266686916351
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,4,1,power_law_1.2,0.23089919090270997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,4,balanced,0.06332266827424367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,4,balanced,0.07720533510049184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,4,balanced,0.11173333724339803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,4,balanced,0.17694934209187826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,4,balanced,0.17851734161376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,4,balanced,0.18196799357732138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,4,balanced,0.18386665980021158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,4,balanced,0.18809600671132407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,4,balanced,0.19131733973821005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,4,balanced,4.96233590443929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,4,1,power_law_1.2,0.40334720611572267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,4,balanced,0.19595199823379517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,4,balanced,0.20332799355189005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,4,balanced,0.21112000942230225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,4,balanced,0.22261333465576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,4,balanced,0.23600000143051147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,4,balanced,0.24765332539876303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,4,balanced,0.27348800500233966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,4,balanced,0.3216000000635783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,4,balanced,0.367301344871521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,4,balanced,0.5325920184453329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,4,balanced,0.6237173477808634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,4,balanced,0.02863999952872594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,4,balanced,0.02934933453798294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,4,balanced,0.03549866626660029
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,4,balanced,0.055946667989095054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,4,balanced,0.07899733384450276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,4,balanced,0.08265066643555959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,4,balanced,0.08521599570910136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,4,balanced,0.08729599912961324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,4,balanced,0.08796800176302592
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,4,balanced,0.08692266543706258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,4,balanced,0.08868267138799031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,4,balanced,0.09050666292508443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,4,balanced,0.09186133742332458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,4,balanced,0.09277866284052531
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,4,balanced,0.12004266182581584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,4,balanced,0.11145599683125813
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,4,balanced,0.10158933202425639
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,4,balanced,0.1823520064353943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,4,balanced,0.1286240021387736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,4,balanced,0.13172800342241922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,4,balanced,0.14408533771832785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,4,balanced,0.19420800606409708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,4,balanced,0.8859466711680094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,4,balanced,0.22187199195226034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,4,balanced,0.31220267216364544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,4,balanced,1.147536039352417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,4,balanced,0.4028373161951701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,4,balanced,0.580069343249003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,4,1,power_law_1.2,0.29189119338989256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,4,balanced,1.6751947402954102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,4,balanced,0.758080005645752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,4,power_law_1.01,0.08131840229034423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,4,1,power_law_1.2,0.5654655933380127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,4,power_law_1.01,0.10382720232009887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,4,balanced,2.1930720011393228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,4,power_law_1.01,0.13919999599456787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,4,power_law_1.01,0.15688320398330688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,4,power_law_1.01,0.16432000398635865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,4,balanced,3.234618822733561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,4,power_law_1.01,0.2223680019378662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,4,power_law_1.01,0.2644351959228516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,4,power_law_1.01,0.2665600061416626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,4,power_law_1.01,0.27268478870391843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,4,power_law_1.01,0.27669761180877683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,4,balanced,4.274431864420573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,4,power_law_1.01,0.2814784049987793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,4,power_law_1.01,0.06945279836654664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,4,power_law_1.01,0.288319993019104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,4,power_law_1.01,0.09259520173072815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,4,power_law_1.01,0.29875199794769286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,4,power_law_1.01,0.10154880285263061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,4,power_law_1.01,0.30726399421691897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,4,power_law_1.01,0.12340480089187622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,4,power_law_1.01,0.32511999607086184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,4,power_law_1.01,0.12861440181732178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,4,power_law_1.01,0.3466048002243042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,4,power_law_1.01,0.15838079452514647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,4,power_law_1.01,0.37080318927764894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,4,power_law_1.01,0.20269439220428467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,4,power_law_1.01,0.20572800636291505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,4,power_law_1.01,0.4656511783599854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,4,power_law_1.01,0.20959360599517823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,4,power_law_1.01,0.5533631801605224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,4,power_law_1.01,0.21157760620117189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,4,power_law_1.01,0.22144639492034912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,4,power_law_1.01,0.6806848049163818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,4,power_law_1.01,0.22836480140686036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,4,power_law_1.01,0.8768383979797363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,4,power_law_1.01,0.23610880374908447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,4,power_law_1.01,0.2593791961669922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,4,power_law_1.01,1.190732765197754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,4,power_law_1.01,0.28981120586395265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,4,power_law_1.01,0.3014591932296753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,4,power_law_1.01,0.3510783910751343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,4,power_law_1.2,1.7192768096923827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,4,power_law_1.01,0.4546624183654785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,4,power_law_1.01,0.4875455856323242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,4,power_law_1.01,0.6848703861236572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,4,power_law_1.01,0.8291007995605468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,4,power_law_1.01,1.5249855995178223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,4,power_law_1.01,1.1279168128967285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,4,power_law_1.2,1.7747711181640624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,4,power_law_1.01,1.5430527687072755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,4,power_law_1.01,2.2475711822509767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,4,1,power_law_1.2,0.7294400215148926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,4,power_law_1.01,2.191097640991211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,4,power_law_1.01,2.8481023788452147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,4,power_law_1.01,2.898361587524414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,4,power_law_1.01,4.380908966064453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,4,power_law_1.01,4.4510337829589846
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,4,1,power_law_1.2,0.39933440685272215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,4,power_law_1.01,6.328185653686523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,4,power_law_1.01,0.06689919829368592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,4,power_law_1.01,0.08804479837417603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,4,power_law_1.01,5.772371292114258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,4,power_law_1.01,0.10080000162124633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,4,power_law_1.01,0.11927039623260498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,4,1,power_law_1.2,1.0112256050109862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,4,power_law_1.01,0.13480960130691527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,4,power_law_1.01,0.16675200462341308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,4,power_law_1.01,0.17662719488143921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,4,power_law_1.01,0.17772799730300903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,4,power_law_1.01,0.18106240034103394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,4,power_law_1.01,0.1829632043838501
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,4,power_law_1.01,0.03129599988460541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,4,power_law_1.01,0.18917759656906127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,4,power_law_1.01,0.041920000314712526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,4,power_law_1.01,0.19788800477981566
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,4,power_law_1.01,0.051744002103805545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,4,power_law_1.01,0.20725760459899903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,4,power_law_1.01,0.05583999752998352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,4,power_law_1.01,0.2149888038635254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,4,power_law_1.01,0.06042879819869995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,4,power_law_1.01,0.24542078971862794
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,4,power_law_1.01,0.0778432011604309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,4,power_law_1.01,0.2568192005157471
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,4,power_law_1.01,0.07950720191001892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,4,power_law_1.01,0.2979135990142822
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,4,power_law_1.01,0.0828224003314972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,4,power_law_1.01,0.3703871965408325
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,4,power_law_1.01,0.08334720134735107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,4,power_law_1.01,0.43511037826538085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,4,power_law_1.01,0.583948802947998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,4,power_law_1.01,0.7225664138793946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,4,power_law_1.01,1.0323712348937988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,4,power_law_1.01,1.3148608207702637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,4,power_law_1.01,0.08471680283546448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,4,power_law_1.01,0.0863103985786438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,4,power_law_1.01,0.08492159843444824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,4,power_law_1.01,1.8872255325317382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,4,power_law_1.01,0.08935040235519409
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,4,power_law_1.01,0.09031680226325989
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,4,power_law_1.01,0.0936959981918335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,4,power_law_1.01,0.10372480154037475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,4,power_law_1.01,2.478118324279785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,4,power_law_1.01,0.10238720178604126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,4,power_law_1.01,0.13347840309143066
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,4,power_law_1.01,0.13129600286483764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,4,power_law_1.01,0.1476032018661499
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,4,power_law_1.01,0.1739583969116211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,4,power_law_1.01,3.961625671386719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,4,power_law_1.01,0.2272320032119751
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,4,power_law_1.01,0.2772543907165527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,4,power_law_1.01,0.38677120208740234
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,4,power_law_1.01,0.5132736206054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,4,power_law_1.01,5.01214714050293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,4,power_law_1.01,0.6820799827575683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,4,power_law_1.2,0.08106880187988282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,4,power_law_1.2,0.09879040122032165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,4,power_law_1.01,1.0538559913635255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,4,power_law_1.2,0.1241536021232605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,4,power_law_1.2,0.16053760051727295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,4,power_law_1.2,0.06721280217170715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,4,power_law_1.2,0.16105600595474243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,4,power_law_1.2,0.08024320006370544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,4,power_law_1.2,0.21816320419311525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,4,power_law_1.2,0.09461119771003723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,4,power_law_1.2,0.266592001914978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,4,power_law_1.2,0.11126400232315063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,4,power_law_1.2,0.26467840671539306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,4,power_law_1.2,0.12362879514694214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,4,power_law_1.2,0.27394559383392336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,4,power_law_1.2,0.1461951971054077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,4,power_law_1.2,0.27670400142669677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,4,power_law_1.2,0.20342400074005126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,4,power_law_1.2,0.28231680393218994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,4,power_law_1.2,0.20832641124725343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,4,power_law_1.2,0.29254400730133057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,4,power_law_1.2,0.2081279993057251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,4,power_law_1.2,0.29617919921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,4,power_law_1.2,0.2146496057510376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,4,power_law_1.2,0.31820800304412844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,4,power_law_1.2,0.22455039024353027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,4,power_law_1.2,0.3313472032546997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,4,power_law_1.2,0.23253118991851807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,4,power_law_1.2,0.3758591890335083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,4,power_law_1.2,0.24419839382171632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,4,power_law_1.2,0.2782655954360962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,4,power_law_1.2,0.40549120903015134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,4,power_law_1.2,0.2915136098861694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,4,power_law_1.2,0.478879976272583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,4,power_law_1.2,0.31493120193481444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,4,power_law_1.2,0.5441023826599121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,4,power_law_1.2,0.7036416053771972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,4,power_law_1.2,0.8844863891601562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,4,power_law_1.2,1.2634559631347657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,4,power_law_1.2,0.3536191940307617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,4,power_law_1.2,0.4301119804382324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,4,power_law_1.2,0.5315392017364502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,4,power_law_1.2,2.1628095626831056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,4,power_law_1.2,0.697753620147705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,4,power_law_1.2,0.8952960014343262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,4,power_law_1.2,1.3157376289367675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,4,power_law_1.2,1.6798080444335937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,4,power_law_1.2,0.06846719980239868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,4,power_law_1.2,0.07288960218429566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,4,power_law_1.2,0.10189440250396728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,4,power_law_1.2,0.11964800357818603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,4,power_law_1.2,0.12420480251312256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,4,power_law_1.2,0.1685439944267273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,4,power_law_1.2,0.17545599937438966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,4,power_law_1.2,0.17727359533309936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,4,power_law_1.2,0.18332159519195557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,4,power_law_1.2,1.5970303535461425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,4,power_law_1.2,0.18532480001449586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,4,power_law_1.2,0.1904255986213684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,4,power_law_1.2,0.19456640481948853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,4,power_law_1.2,0.21524479389190673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,4,power_law_1.2,0.23440001010894776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,4,power_law_1.2,0.2520639896392822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,4,power_law_1.2,0.26702721118927003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,4,power_law_1.2,0.2997119903564453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,4,power_law_1.2,0.3745919942855835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,4,power_law_1.2,0.4428800106048584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,4,power_law_1.2,0.5925119876861572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,4,power_law_1.2,0.7427199840545654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,4,power_law_1.2,1.088268756866455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,4,power_law_1.2,2.3534912109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,4,power_law_1.2,1.4005632400512695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,4,1,power_law_1.2,1.3229311943054198
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,4,power_law_1.2,0.3766592025756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,4,power_law_1.2,1.95533447265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,4,power_law_1.2,0.03293440043926239
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,4,power_law_1.2,0.03818239867687225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,4,power_law_1.2,0.05273600220680237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,4,power_law_1.2,0.05486080050468445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,4,power_law_1.2,2.239142417907715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,4,1,balanced,0.04350399971008301
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,4,power_law_1.2,0.06063359975814819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,4,1,balanced,0.06011733412742615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,4,1,balanced,0.07658133407433827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,4,1,balanced,0.12889599800109863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,4,1,balanced,0.238864004611969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,4,1,balanced,0.2386666735013326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,4,1,balanced,0.23797865708669028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,4,1,balanced,0.23879466454188028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,4,1,balanced,0.24064532915751138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,4,1,balanced,0.24328533808390299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,4,1,balanced,0.24147733052571616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,4,1,balanced,0.26876266797383624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,4,1,balanced,0.2711893320083618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,4,1,balanced,0.27663467327753705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,4,1,balanced,0.28384000062942505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,4,power_law_1.2,3.210521697998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,4,1,balanced,0.2903253237406413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,4,1,balanced,0.30352532863616943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,4,1,balanced,0.32846399148305255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,4,1,balanced,0.3577973445256551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,4,1,balanced,0.46730132897694904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,4,1,balanced,0.5301013390223185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,4,1,balanced,0.7366027037302653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,4,1,balanced,0.9329173564910889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,4,1,balanced,1.3508000373840332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,4,1,balanced,1.7531572977701824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,4,1,power_law_1.2,1.0578368186950684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,4,power_law_1.2,0.0765504002571106
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,4,power_law_1.2,0.07978240251541138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,4,power_law_1.2,0.08370559811592101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,4,1,balanced,2.6159253120422363
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,4,power_law_1.2,0.08316799998283386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,4,power_law_1.2,0.08376960158348083
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,4,power_law_1.2,0.0866752028465271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,4,1,balanced,3.4088265101114907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,4,1,balanced,0.037434667348861694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,4,1,balanced,0.052058666944503784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,4,1,balanced,0.07248533268769582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,4,1,balanced,0.12173333764076233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,4,1,power_law_1.2,0.5236095905303955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,4,1,balanced,0.18684800465901694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,4,1,balanced,0.18901866674423218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,4,1,balanced,0.18922666708628336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,4,1,balanced,0.19197867314020792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,4,power_law_1.2,2.453299140930176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,4,1,balanced,0.19199466705322266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,4,1,balanced,0.19260799884796143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,4,1,balanced,0.19267199436823526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,4,1,balanced,0.037658666570981346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,4,1,balanced,0.21252266565958658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,4,1,balanced,0.04762666424115499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,4,1,balanced,0.21970667441685995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,4,1,balanced,0.0621973325808843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,4,power_law_1.2,2.6776384353637694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,4,1,balanced,0.09118933478991191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,4,1,balanced,0.15990933775901794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,4,1,balanced,0.15533333023389181
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,4,power_law_1.2,0.08373759984970093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,4,1,balanced,0.1569386621316274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,4,1,balanced,0.15852800011634827
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,4,power_law_1.2,0.09036800265312195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,4,power_law_1.2,0.09365119934082031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,4,power_law_1.2,0.09944319725036621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,4,power_law_1.2,0.10140160322189332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,4,power_law_1.2,0.10707199573516846
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,4,power_law_1.2,0.15555200576782227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,4,1,balanced,0.2258346676826477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,4,1,balanced,0.23085333903630575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,4,1,balanced,0.2394719918568929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,4,1,balanced,0.25380265712738037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,4,1,balanced,0.2836640079816182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,4,1,balanced,0.15773866573969522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,4,1,balanced,0.16009066502253214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,4,power_law_1.2,3.166617584228516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,4,1,balanced,0.1646613379319509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,4,1,balanced,0.16581333676973978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,4,1,balanced,0.1684373418490092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,4,1,balanced,0.17890133460362753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,4,1,balanced,0.22616533438364664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,4,1,balanced,0.23281600077946982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,4,1,balanced,0.23703465859095255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,4,1,balanced,0.25456533829371136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,4,1,balanced,0.27876800298690796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,4,1,balanced,0.42421332995096844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,4,power_law_1.2,0.15900800228118897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,4,1,balanced,0.4642133315404256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,4,power_law_1.2,0.1584895968437195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,4,1,balanced,0.3168906569480896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,4,1,balanced,0.6502133210500082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,4,power_law_1.2,0.18455679416656495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,4,1,balanced,0.4580213228861491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,4,1,balanced,0.8469706376393636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,4,power_law_1.2,0.2596927881240845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,4,1,balanced,1.2301973501841228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,4,1,balanced,1.6263200441996257
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,4,power_law_1.2,0.2815808057785034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,4,1,balanced,2.3831307093302407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,4,1,balanced,0.5203786691029867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,4,1,balanced,0.7358506520589193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,4,1,balanced,3.153034528096517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,4,1,balanced,0.9435679912567139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,4,1,balanced,1.3649439811706543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,4,1,balanced,1.7886773745218914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,4,1,balanced,2.6127039591471353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,4,1,balanced,3.4524428049723306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,4,1,power_law_1.01,0.034745600819587705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,4,1,balanced,0.027503999571005504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,4,1,balanced,0.02942933390537898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,4,1,power_law_1.01,0.057132798433303836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,4,1,balanced,0.03589333345492681
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,4,1,balanced,0.05599466462930044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,4,1,power_law_1.01,0.07640960216522216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,4,1,balanced,0.08106666803359985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,4,1,balanced,0.08272533118724823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,4,1,power_law_1.01,0.108787202835083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,4,1,power_law_1.01,0.13304959535598754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,4,1,power_law_1.01,0.21247999668121337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,4,1,power_law_1.01,0.21778559684753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,4,1,power_law_1.01,0.2290560007095337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,4,1,power_law_1.01,0.2318592071533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,4,1,power_law_1.01,0.2320768117904663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,4,1,power_law_1.01,0.2502016067504883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,4,1,balanced,0.08658132950464885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,4,1,power_law_1.01,0.26049280166625977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,4,1,balanced,0.08737599849700928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,4,1,balanced,0.09077333410580952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,4,1,power_law_1.01,0.2648767948150635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,4,1,balanced,0.09230933586756389
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,4,1,balanced,0.09478933612505595
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,4,1,balanced,0.10454932848612468
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,4,1,balanced,0.10748799641927083
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,4,1,balanced,0.11492799719174702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,4,1,balanced,0.1134986678759257
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,4,1,balanced,0.1153600017229716
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,4,1,balanced,0.1251200040181478
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,4,1,balanced,0.13223999738693237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,4,1,balanced,0.14518933494885763
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,4,1,balanced,0.18195732434590658
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,4,1,balanced,0.1853920022646586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,4,1,balanced,0.35764801502227783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,4,1,balanced,0.2741386691729228
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,4,1,balanced,0.383407990137736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,4,power_law_1.2,3.873062515258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,4,1,power_law_1.01,0.2726399898529053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,4,1,power_law_1.01,0.27903358936309813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,4,1,power_law_1.01,0.2870527982711792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,4,power_law_1.2,0.39335041046142577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,4,1,power_law_1.01,0.3123584032058716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,4,1,power_law_1.01,0.370412802696228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,4,1,power_law_1.01,0.40558719635009766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,4,1,power_law_1.01,0.5071743965148926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,4,1,balanced,0.49107201894124347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,4,1,power_law_1.01,0.5912447929382324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,4,1,balanced,0.6999733448028564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,4,1,power_law_1.01,0.8017215728759766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,4,1,balanced,0.9144000212351481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,4,1,power_law_1.01,0.030393600463867188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,4,1,power_law_1.01,1.0083840370178223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,4,1,power_law_1.01,0.05041279792785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,4,1,power_law_1.01,0.06711680293083191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,4,1,power_law_1.01,1.4103551864624024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,4,1,power_law_1.01,0.09274880290031433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,4,1,power_law_1.01,0.11742080450057983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,4,1,power_law_1.01,0.1525887966156006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,4,1,power_law_1.01,1.8380800247192384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,4,1,power_law_1.01,0.15678720474243163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,4,1,power_law_1.01,0.16389119625091553
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,4,power_law_1.2,0.46222720146179197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,4,1,power_law_1.01,0.181113600730896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,4,1,power_law_1.01,0.18873599767684937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,4,1,power_law_1.01,2.6730815887451174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,4,1,power_law_1.01,0.20305280685424804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,4,1,power_law_1.01,0.20843520164489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,4,1,power_law_1.01,0.03669120073318481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,4,1,power_law_1.01,0.2145983934402466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,4,1,power_law_1.01,0.044998401403427125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,4,1,power_law_1.01,0.22669439315795897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,4,1,power_law_1.01,0.061324799060821535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,4,1,power_law_1.01,0.23950719833374023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,4,1,power_law_1.01,0.07752959728240967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,4,1,power_law_1.01,3.5179775238037108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,4,1,power_law_1.01,0.25052800178527834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,4,1,power_law_1.01,0.09413759708404541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,4,1,power_law_1.01,0.2776576042175293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,4,1,power_law_1.01,0.14275840520858765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,4,1,power_law_1.01,0.32737278938293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,4,1,power_law_1.01,0.1444416046142578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,4,1,power_law_1.01,0.14693119525909423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,4,1,power_law_1.01,0.15282560586929322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,4,1,power_law_1.01,0.1549183964729309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,4,1,power_law_1.01,0.16979199647903442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,4,1,power_law_1.01,0.18368639945983886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,4,1,power_law_1.01,0.1894912004470825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,4,1,power_law_1.01,0.19487359523773193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,4,1,power_law_1.01,0.2120512008666992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,4,1,power_law_1.01,0.3849087953567505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,4,1,power_law_1.01,0.23415679931640626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,4,1,power_law_1.01,0.026579201221466064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,4,1,power_law_1.01,0.48349437713623045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,4,1,power_law_1.01,0.02913280129432678
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,4,1,power_law_1.01,0.03991039991378784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,4,1,power_law_1.01,0.5872384071350097
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,4,1,power_law_1.01,0.04791040122509003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,4,1,power_law_1.01,0.06023039817810059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,4,1,power_law_1.01,0.7897344112396241
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,4,1,power_law_1.01,0.07496320009231568
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,4,1,power_law_1.01,0.07882239818572997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,4,1,power_law_1.01,0.9945792198181153
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,4,1,power_law_1.01,0.08451200127601624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,4,1,power_law_1.01,0.08718720078468323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,4,1,power_law_1.01,0.25511679649353025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,4,1,power_law_1.01,1.4107135772705077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,4,1,power_law_1.01,0.2889535903930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,4,1,power_law_1.01,0.34610559940338137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,4,1,power_law_1.01,1.8243904113769531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,4,1,power_law_1.01,0.43396477699279784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,4,1,power_law_1.01,0.5252287864685059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,4,1,power_law_1.01,0.7215360164642334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,4,1,power_law_1.01,2.6631296157836912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,4,1,power_law_1.01,0.9112895965576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,4,1,power_law_1.01,1.2937984466552734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,4,1,power_law_1.01,3.4880126953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,4,1,power_law_1.2,0.03505280017852783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,4,1,power_law_1.01,1.6763200759887695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,4,1,power_law_1.2,0.05692800283432007
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,4,1,power_law_1.01,0.08746240139007569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,4,1,power_law_1.2,0.07436800003051758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,4,1,power_law_1.01,0.09423999786376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,4,1,power_law_1.2,0.09903360009193421
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,4,1,power_law_1.01,0.10033279657363892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,4,1,power_law_1.2,0.12232320308685303
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,4,1,power_law_1.01,0.1049407958984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,4,1,power_law_1.01,2.4410751342773436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,4,1,power_law_1.2,0.21245439052581788
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,4,1,power_law_1.01,0.1164736032485962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,4,1,power_law_1.2,0.21747839450836182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,4,1,power_law_1.01,0.10767359733581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,4,1,power_law_1.2,0.22524800300598144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,4,1,power_law_1.2,0.2294975996017456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,4,power_law_1.2,4.655180740356445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,4,1,power_law_1.2,0.23384320735931396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,4,1,power_law_1.01,3.2001792907714846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,4,1,power_law_1.2,0.25635199546813964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,4,1,power_law_1.2,0.03333120048046112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,4,1,power_law_1.2,0.2610176086425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,4,1,power_law_1.2,0.051283198595047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,4,1,power_law_1.2,0.26547839641571047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,4,1,power_law_1.2,0.06618239879608154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,4,1,power_law_1.2,0.27112319469451907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,4,1,power_law_1.2,0.08806399703025818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,4,1,power_law_1.2,0.2758271932601929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,4,1,power_law_1.2,0.10740480422973633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,4,1,power_law_1.2,0.15219199657440186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,4,1,power_law_1.2,0.16252800226211547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,4,1,power_law_1.2,0.17667839527130128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,4,1,power_law_1.2,0.18250240087509156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,4,1,power_law_1.2,0.19332480430603027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,4,power_law_1.2,4.959910583496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,4,1,power_law_1.2,0.20921599864959717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,4,1,power_law_1.2,0.2132416009902954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,4,1,power_law_1.01,0.11315200328826905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,4,1,power_law_1.2,0.28482561111450194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,4,1,power_law_1.2,0.3100480079650879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,4,1,power_law_1.2,0.371014404296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,4,1,power_law_1.2,0.4069568157196045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,4,1,power_law_1.2,0.5134655952453613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,4,1,power_law_1.2,0.6067584037780762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,4,1,power_law_1.2,0.22062079906463622
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,4,1,power_law_1.01,0.12832000255584716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,4,1,power_law_1.2,0.23084800243377684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,4,1,power_law_1.01,0.13264000415802002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,4,1,power_law_1.2,0.24097919464111328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,4,1,power_law_1.01,0.1522752046585083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,4,1,power_law_1.2,0.2550015926361084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,4,1,power_law_1.01,0.16627199649810792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,4,1,power_law_1.2,0.2858304023742676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,4,1,power_law_1.01,0.19576319456100463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,4,1,power_law_1.2,0.33827838897705076
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,4,1,power_law_1.01,0.24588799476623535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,4,1,power_law_1.2,0.8078144073486329
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,4,1,power_law_1.01,0.29656319618225097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,4,1,power_law_1.2,1.0135104179382324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,4,1,power_law_1.01,0.4087808132171631
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,4,1,power_law_1.01,0.5188543796539307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,4,1,power_law_1.01,0.7297279834747314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,4,1,power_law_1.2,0.3846271991729736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,4,1,power_law_1.2,0.48932480812072754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,4,1,power_law_1.2,0.5902592182159424
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,4,1,power_law_1.01,0.9966272354125977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,4,1,power_law_1.2,0.03641600012779236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,4,1,power_law_1.2,0.04496000111103058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,4,1,power_law_1.2,0.060172802209854125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,4,1,power_law_1.2,0.07367039918899536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,4,1,power_law_1.2,0.08632320165634155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,4,1,power_law_1.2,0.7933119773864746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,4,1,power_law_1.2,1.4285247802734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,4,power_law_1.2,5.407897567749023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,4,1,power_law_1.2,1.0005503654479981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,4,1,power_law_1.2,0.14108799695968627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,4,power_law_1.2,0.5383423805236817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,4,1,power_law_1.2,0.14648959636688233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,4,1,power_law_1.2,0.14817919731140136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,4,1,power_law_1.2,0.15223679542541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,4,1,power_law_1.2,0.15557119846343995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,4,1,power_law_1.2,0.17345279455184937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,4,1,power_law_1.2,0.18465280532836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,4,1,power_law_1.2,0.19561599493026732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,4,1,power_law_1.2,0.19540480375289918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,4,1,power_law_1.2,0.026950401067733765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,4,1,power_law_1.2,0.21442561149597167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,4,1,power_law_1.2,0.029721599817276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,4,1,power_law_1.2,0.2415679931640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,4,1,power_law_1.2,0.03621760010719299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,4,1,power_law_1.2,0.2624768018722534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,4,1,power_law_1.2,1.4184320449829102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,4,1,power_law_1.2,0.045414400100708005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,4,1,power_law_1.2,0.29431040287017823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,4,1,power_law_1.2,0.058310401439666745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,4,1,power_law_1.2,0.34719998836517335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,4,1,power_law_1.2,0.07441920042037964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,4,1,power_law_1.2,0.43534078598022463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,4,1,power_law_1.2,0.07887359857559204
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,4,1,power_law_1.2,0.08420479893684388
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,4,1,power_law_1.2,0.08668799996376038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,4,1,power_law_1.2,0.5384511947631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,4,1,power_law_1.2,0.08869760036468506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,4,1,power_law_1.2,0.09437440037727356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,4,1,power_law_1.2,0.10016000270843506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,4,1,power_law_1.2,0.106278395652771
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,4,1,power_law_1.2,0.11782399415969849
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,4,1,power_law_1.2,0.10958720445632934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,4,1,power_law_1.2,0.11461119651794434
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,4,1,power_law_1.2,0.1292736053466797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,4,1,power_law_1.2,0.13463040590286254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,4,1,power_law_1.2,0.7232895851135254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,4,1,power_law_1.2,0.9153984069824219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,4,1,power_law_1.2,0.155404794216156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,4,1,power_law_1.2,0.16855039596557617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,4,1,power_law_1.2,0.19619840383529663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,4,power_law_1.2,6.558662414550781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,4,balanced,0.08826133608818054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,4,balanced,0.10739200313886006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,4,balanced,0.16406400005022684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,4,1,power_law_1.2,0.24652159214019775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,4,balanced,0.3237386743227641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,4,balanced,0.5309013525644938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,4,1,power_law_1.2,1.2937536239624023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,4,balanced,0.6701333522796631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,4,balanced,0.7990026473999023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,4,balanced,0.7821919918060303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,4,balanced,0.7754027048746744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,4,1,power_law_1.2,1.8402559280395507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,4,balanced,0.7900746663411459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,4,balanced,0.8193173408508301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,4,balanced,0.8209013144175211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,4,1,power_law_1.2,0.30332159996032715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,4,balanced,0.8278186321258545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,4,balanced,0.8311200141906738
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,4,power_law_1.2,0.7190783977508545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,4,balanced,0.8420053323109945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,4,balanced,0.8484319845835367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,4,balanced,0.8734719753265381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,4,balanced,0.9283146858215332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,4,balanced,0.9989706675211588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,4,balanced,1.4070347150166829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,4,balanced,1.564901351928711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,4,1,power_law_1.2,1.835865592956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,4,balanced,2.2586560249328613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,4,balanced,2.85150941212972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,4,balanced,3.961973190307617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,4,balanced,5.841877619425456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,4,1,power_law_1.2,0.41182718276977537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,4,1,power_law_1.2,1.6718975067138673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,4,balanced,9.153562545776367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,4,1,power_law_1.2,0.5262656211853027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,4,balanced,12.571829477945963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,4,balanced,0.07955199976762135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,4,balanced,0.09270399808883667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,4,balanced,0.13531733552614847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,4,balanced,0.20428800582885742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,4,balanced,0.3534506559371948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,4,balanced,0.43403200308481854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,4,balanced,0.4790453513463338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,4,balanced,0.4833600123723348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,4,balanced,0.48869868119557697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,4,balanced,0.492298682530721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,4,balanced,0.5029600063959757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,4,balanced,0.512165347735087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,4,balanced,0.5207253297170004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,4,balanced,0.5394719839096069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,4,balanced,0.558512012163798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,4,balanced,0.5889813502629598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,4,balanced,0.6337706645329794
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,4,1,power_law_1.2,0.7327744007110596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,4,balanced,0.7283626397450765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,4,balanced,0.8240266640981039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,4,balanced,1.2357333501180012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,4,balanced,1.414415995279948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,4,balanced,2.033082644144694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,4,balanced,2.6462507247924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,4,balanced,3.6429920196533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,4,balanced,4.868656158447266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,4,balanced,7.168080012003581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,4,balanced,10.497642517089844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,4,balanced,0.06705600023269653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,4,balanced,0.09230400125185649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,4,balanced,0.10566400488217671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,4,balanced,0.1767680048942566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,4,balanced,0.30666667222976685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,4,balanced,0.35550932089487713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,4,balanced,0.35859731833140057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,4,balanced,0.3656746546427409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,4,balanced,0.37305064996083576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,4,balanced,0.4187146822611491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,4,balanced,0.42985065778096515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,4,balanced,0.43566401799519855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,4,balanced,0.44334932168324787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,4,balanced,0.48418664932250977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,4,balanced,0.503167986869812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,4,balanced,0.5218079884847006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,4,balanced,0.5824906826019287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,4,balanced,0.6506400108337402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,4,balanced,0.6917226314544678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,4,balanced,1.0524213314056396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,4,balanced,1.1613279978434246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,4,balanced,1.6772267023722331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,4,1,power_law_1.2,2.6723968505859377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,4,balanced,2.1698452631632485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,4,balanced,2.939765294392904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,4,balanced,3.933055877685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,4,1,power_law_1.2,2.437446403503418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,4,power_law_1.2,5.917452621459961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,4,balanced,5.6829172770182295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,4,1,power_law_1.2,1.385747241973877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,4,power_law_1.01,0.15008000135421753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,4,power_law_1.01,0.28843519687652586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,4,balanced,7.432767868041992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,4,power_law_1.01,0.30275840759277345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,4,power_law_1.01,0.38657920360565184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,4,power_law_1.01,0.1238592028617859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,4,power_law_1.01,0.47940478324890134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,4,power_law_1.01,0.19262720346450807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,4,power_law_1.01,0.2081216096878052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,4,1,power_law_1.2,2.6661439895629884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,4,power_law_1.01,0.26830079555511477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,4,power_law_1.01,0.25374081134796145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,4,power_law_1.01,0.39373440742492677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,4,power_law_1.01,0.4324607849121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,4,power_law_1.01,0.4512063980102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,4,power_law_1.01,0.5338431835174561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,4,power_law_1.01,0.4694784164428711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,4,power_law_1.01,0.46983041763305666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,4,power_law_1.01,0.766425609588623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,4,power_law_1.01,0.48974080085754396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,4,balanced,0.0315786674618721
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,4,balanced,0.041477332512537636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,4,balanced,0.06214400132497152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,4,power_law_1.01,0.508019208908081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,4,power_law_1.01,0.5197760105133057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,4,power_law_1.01,0.549126386642456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,4,power_law_1.01,0.5922560214996337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,4,power_law_1.01,0.6368063926696778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,4,power_law_1.01,0.7119743824005127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,4,power_law_1.01,0.6950463771820068
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,4,balanced,0.09151466687520345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,4,balanced,0.14945066968599954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,4,power_law_1.01,0.7012095928192139
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,4,balanced,0.1806559960047404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,4,balanced,0.1835040052731832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,4,balanced,0.1885333259900411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,4,power_law_1.01,0.7353151798248291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,4,power_law_1.01,0.6961472034454346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,4,power_law_1.01,0.7691391944885254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,4,power_law_1.01,0.7947648048400879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,4,power_law_1.01,0.8183103561401367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,4,power_law_1.01,0.8489343643188476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,4,balanced,0.19194666544596353
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,4,balanced,0.1876373291015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,4,power_law_1.01,0.8692159652709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,4,balanced,0.19100799163182577
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,4,balanced,0.19373865922292074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,4,balanced,0.19201600551605225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,4,power_law_1.01,0.8757247924804688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,4,balanced,0.19921600818634033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,4,balanced,0.20360000928243002
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,4,balanced,0.26613332827885944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,4,power_law_1.01,0.9196096420288086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,4,power_law_1.01,1.0712320327758789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,4,power_law_1.01,1.207545566558838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,4,power_law_1.01,1.0211071968078613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,4,power_law_1.01,1.5623807907104492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,4,power_law_1.01,1.3546751976013183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,4,balanced,0.2398773431777954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,4,balanced,0.5099999904632568
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,4,balanced,0.37830932935078937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,4,balanced,0.2602506677309672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,4,balanced,0.2792746623357137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,4,balanced,0.406335989634196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,4,balanced,0.44018133481343585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,4,balanced,0.628058671951294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,4,balanced,0.8166293303171793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,4,balanced,1.0848426818847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,4,power_law_1.01,1.7207168579101562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,4,balanced,1.6462079683939617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,4,1,power_law_1.2,3.4952831268310547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,4,power_law_1.01,2.439967918395996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,4,power_law_1.01,0.10030080080032348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,4,power_law_1.01,0.15733760595321655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,4,power_law_1.01,1.8660991668701172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,4,power_law_1.01,3.0209024429321287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,4,power_law_1.01,2.542803192138672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,4,power_law_1.01,4.209740829467774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,4,power_law_1.01,0.17634559869766236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,4,power_law_1.01,3.2866752624511717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,4,power_law_1.01,0.19372160434722902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,4,power_law_1.01,0.24959359169006348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,4,power_law_1.01,0.30215680599212646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,4,power_law_1.01,0.32391679286956787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,4,power_law_1.01,5.547872161865234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,4,power_law_1.01,0.34613120555877686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,4,power_law_1.01,4.616966247558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,4,power_law_1.01,0.3363071918487549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,4,power_law_1.01,0.3897279977798462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,4,power_law_1.01,0.3992831945419312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,4,power_law_1.01,0.06053119897842407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,4,power_law_1.01,0.4144768238067627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,4,power_law_1.01,0.08156160116195679
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,4,power_law_1.01,0.0810368001461029
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,4,power_law_1.01,0.1081279993057251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,4,power_law_1.01,5.922835159301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,4,power_law_1.01,0.12821760177612304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,4,power_law_1.01,8.670214080810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,4,power_law_1.01,0.15882240533828734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,4,power_law_1.01,0.17089920043945311
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,4,power_law_1.01,0.1678015947341919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,4,power_law_1.01,0.1814463973045349
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,4,power_law_1.01,0.17667200565338134
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,4,power_law_1.01,0.18600319623947142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,4,power_law_1.01,0.18426239490509033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,4,power_law_1.01,0.18852479457855226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,4,power_law_1.01,0.19507839679718017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,4,power_law_1.01,0.20821759700775147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,4,power_law_1.01,9.158528137207032
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,4,power_law_1.01,0.21660161018371582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,4,power_law_1.01,0.22303359508514403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,4,power_law_1.01,0.4580992221832275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,4,power_law_1.01,0.31146879196166993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,4,power_law_1.01,0.4814591884613037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,4,power_law_1.01,11.222905731201172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,4,power_law_1.01,0.31699841022491454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,4,power_law_1.01,0.5081791877746582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,4,power_law_1.2,0.1512063980102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,4,power_law_1.01,0.3232640027999878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,4,power_law_1.01,0.5320320129394531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,4,power_law_1.2,0.21881599426269532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,4,power_law_1.01,0.3662463903427124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,4,power_law_1.01,0.6256447792053222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,4,power_law_1.2,0.3097856044769287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,4,power_law_1.01,0.46761598587036135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,4,power_law_1.2,0.321612811088562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,4,power_law_1.01,0.7260672092437744
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,4,power_law_1.01,0.5616447925567627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,4,power_law_1.2,0.3589312076568604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,4,power_law_1.01,0.8253696441650391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,4,power_law_1.2,0.5484479904174805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,4,power_law_1.01,0.7621056079864502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,4,power_law_1.2,0.6578815937042236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,4,power_law_1.01,1.0950528144836427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,4,power_law_1.01,0.9605183601379395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,4,power_law_1.2,0.7285120010375976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,4,power_law_1.01,12.487359619140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,4,power_law_1.2,0.7048064231872558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,4,power_law_1.01,1.369318389892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,4,power_law_1.01,1.353708839416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,4,power_law_1.2,0.13093119859695435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,4,power_law_1.2,0.707590389251709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,4,power_law_1.2,0.16801919937133789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,4,power_law_1.2,0.7921728134155274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,4,power_law_1.01,1.9053184509277343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,4,power_law_1.2,0.1956544041633606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,4,power_law_1.2,0.2523519992828369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,4,power_law_1.2,0.8048895835876465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,4,power_law_1.01,2.089523124694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,4,power_law_1.2,0.26423680782318115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,4,power_law_1.2,0.7986048221588135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,4,power_law_1.01,2.4404415130615233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,4,power_law_1.2,0.8430656433105469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,4,1,power_law_1.2,3.508313751220703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,4,power_law_1.2,0.848902416229248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,4,power_law_1.2,0.10024319887161255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,4,power_law_1.2,0.8971712112426757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,4,power_law_1.01,3.3589759826660157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,4,power_law_1.2,0.9693951606750488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,4,power_law_1.2,0.39201281070709226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,4,power_law_1.2,1.082425594329834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,4,power_law_1.2,0.4408256053924561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,4,power_law_1.2,0.4454592227935791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,4,power_law_1.2,1.3128576278686523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,4,power_law_1.2,0.45865597724914553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,4,power_law_1.01,4.383718490600586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,4,power_law_1.2,0.47160959243774414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,4,power_law_1.2,1.6546239852905273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,4,power_law_1.2,0.5016511917114258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,4,power_law_1.2,0.5078015804290772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,4,power_law_1.2,1.933024024963379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,4,power_law_1.2,0.5319424152374268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,4,power_law_1.2,0.5933568000793457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,4,power_law_1.2,0.5999167919158935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,4,power_law_1.2,2.7558208465576173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,4,power_law_1.01,6.610111999511719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,4,power_law_1.2,0.6343743801116943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,4,power_law_1.2,0.7375872135162354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,4,power_law_1.2,0.8862400054931641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,4,power_law_1.2,3.501996612548828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,4,power_law_1.2,0.058963197469711306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,4,power_law_1.2,0.06551039814949036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,4,power_law_1.2,1.052076816558838
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,4,power_law_1.2,0.07655680179595947
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,4,power_law_1.2,0.11427839994430541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,4,power_law_1.2,1.45217924118042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,4,power_law_1.2,0.1160256028175354
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,4,power_law_1.2,0.1621567964553833
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,4,power_law_1.2,0.15962239503860473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,4,power_law_1.2,1.8381183624267579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,4,power_law_1.01,8.85754852294922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,4,power_law_1.2,4.980633544921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,4,power_law_1.2,0.1681663990020752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,4,1,balanced,0.07424533367156982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,4,1,balanced,0.09459732969601949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,4,1,balanced,0.14613333344459534
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,4,power_law_1.2,0.17863680124282838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,4,1,balanced,0.3177280028661092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,4,power_law_1.2,0.1780992031097412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,4,power_law_1.2,2.381875228881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,4,power_law_1.2,6.73565444946289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,4,power_law_1.2,3.291334533691406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,4,1,balanced,0.5171840190887451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,4,1,balanced,0.6593653361002604
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,4,power_law_1.2,0.18232959508895874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,4,1,balanced,0.6605439980824789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,4,power_law_1.2,0.18802560567855836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,4,1,balanced,0.6621439854303995
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,4,power_law_1.2,0.1872383952140808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,4,1,balanced,0.6615999937057495
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,4,power_law_1.2,0.20154240131378173
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,4,power_law_1.2,0.20651519298553467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,4,power_law_1.2,0.22650880813598634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,4,power_law_1.2,4.508089447021485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,4,power_law_1.2,0.2329535961151123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,4,power_law_1.2,0.3734080076217651
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,4,power_law_1.2,0.31255040168762205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,4,power_law_1.2,0.35729279518127444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,4,1,balanced,0.6624533335367838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,4,1,balanced,0.667248010635376
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,4,power_law_1.2,0.41340160369873047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,4,1,balanced,0.6696533362070719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,4,power_law_1.2,0.473529577255249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,4,1,balanced,0.7373706499735514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,4,1,balanced,0.7598826885223389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,4,power_law_1.2,6.184601593017578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,4,power_law_1.2,0.6329984188079834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,4,1,balanced,0.7789173126220703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,4,1,balanced,0.7829439640045166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,4,power_law_1.2,0.8570560455322266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,4,1,balanced,0.7784586747487386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,4,1,balanced,0.8149440288543701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,4,power_law_1.2,0.12151679992675782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,4,1,balanced,0.8689173062642416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,4,power_law_1.2,9.852460479736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,4,power_law_1.2,0.16421760320663453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,4,1,balanced,1.224778652191162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,4,power_law_1.2,0.20727040767669677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,4,1,balanced,1.318885326385498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,4,power_law_1.2,0.22923519611358642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,4,power_law_1.2,0.29762558937072753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,4,1,balanced,1.9523839950561523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,4,power_law_1.2,0.3302848100662231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,4,power_law_1.2,0.33783679008483886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,4,power_law_1.2,9.795334625244141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,4,1,balanced,2.438368002573649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,4,power_law_1.2,0.3447423934936523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,4,1,power_law_1.2,3.1976255416870116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,4,1,balanced,0.06607466439406078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,4,1,balanced,3.105989456176758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,4,1,balanced,0.08480000495910645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,4,1,balanced,0.11598933736483256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,4,1,balanced,0.20896534125010172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,4,1,balanced,0.37215999762217206
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,4,power_law_1.2,0.9435647964477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,4,1,balanced,0.45582934220631915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,4,1,balanced,0.45741868019104004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,4,1,balanced,0.4578933318456014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,4,1,balanced,4.323370615641276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,4,power_law_1.2,1.545740795135498
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,4,power_law_1.2,2.331110382080078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,4,1,power_law_1.2,1.0177663803100585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,4,1,balanced,6.131327946980794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,4,1,balanced,0.45949868361155194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,4,1,balanced,0.4604426622390747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,4,1,balanced,0.05612266560395559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,4,1,balanced,0.07286933561166127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,4,1,balanced,0.10276800394058228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,4,1,balanced,0.18392000595728555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,4,1,balanced,0.3062079946200053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,4,1,balanced,0.3749759991963704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,4,1,balanced,0.3574613332748413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,4,1,balanced,0.3567253351211548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,4,1,balanced,0.3598080078760783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,4,1,balanced,0.3589706818262736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,4,1,balanced,0.36694931983947754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,4,1,balanced,8.292533238728842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,4,1,balanced,0.369978666305542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,4,1,balanced,0.37172798315684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,4,1,power_law_1.01,0.08035200238227844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,4,1,balanced,0.3815893332163493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,4,1,balanced,0.4655839999516805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,4,1,power_law_1.01,0.09758080244064331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,4,1,balanced,0.4705919822057088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,4,1,power_law_1.01,0.1487936019897461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,4,1,power_law_1.01,0.28465919494628905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,4,1,power_law_1.01,0.33253118991851804
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,4,power_law_1.2,1.080851173400879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,4,1,power_law_1.01,0.5070144176483155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,4,1,balanced,0.032058666149775185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,4,1,power_law_1.01,0.06410239934921265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,4,1,power_law_1.01,0.5344575881958008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,4,1,power_law_1.01,0.08047360181808472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,4,1,balanced,0.3908373514811198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,4,1,power_law_1.01,0.1136896014213562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,4,1,power_law_1.01,0.5643392086029053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,4,1,balanced,0.4652640024820964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,4,1,power_law_1.01,0.18465280532836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,4,power_law_1.2,11.854547119140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,4,1,power_law_1.01,0.5611199855804443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,4,1,power_law_1.01,0.24248960018157958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,4,1,power_law_1.01,0.602623987197876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,4,1,power_law_1.01,0.3647808074951172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,4,1,power_law_1.01,0.6182655811309814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,4,1,power_law_1.01,0.3852992057800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,4,1,balanced,0.04368533194065094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,4,1,power_law_1.01,0.6729536056518555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,4,1,power_law_1.01,0.39296000003814696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,4,1,power_law_1.01,0.053311997652053834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,4,1,power_law_1.01,0.3948735952377319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,4,1,power_law_1.01,0.6929279804229737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,4,1,power_law_1.01,0.0719871997833252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,4,1,balanced,0.5460213422775269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,4,1,power_law_1.01,0.4291840076446533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,4,1,power_law_1.01,0.09873279929161072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,4,1,balanced,0.49650665124257404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,4,1,power_law_1.01,0.7476672172546387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,4,1,power_law_1.01,0.47896318435668944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,4,1,balanced,0.8587253093719482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,4,1,power_law_1.01,0.7630911827087402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,4,1,balanced,0.6099626620610555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,4,1,power_law_1.01,0.5047743797302247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,4,1,balanced,0.9336533546447754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,4,1,power_law_1.01,0.45223679542541506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,4,1,balanced,1.0054826736450195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,4,1,power_law_1.01,0.473145580291748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,4,1,power_law_1.01,0.5119423866271973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,4,1,balanced,1.443413257598877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,4,1,power_law_1.01,0.5395391941070556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,4,1,balanced,0.47712000211079914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,4,1,balanced,1.8913812637329102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,4,1,balanced,0.48981332778930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,4,1,power_law_1.01,0.5801280021667481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,4,1,balanced,0.5061440070470175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,4,power_law_1.2,12.938803100585938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,4,1,balanced,0.5323786735534668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,4,1,power_law_1.01,0.6840576171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,4,1,balanced,0.5973920027414957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,4,1,balanced,2.4663306872049966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,4,1,power_law_1.01,0.7786496162414551
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,4,1,balanced,0.0646613339583079
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,4,1,balanced,0.094842662413915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,4,1,balanced,0.1558133363723755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,4,1,balanced,0.18695465723673502
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,4,1,balanced,0.19035200277964273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,4,1,balanced,0.1974560022354126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,4,1,balanced,3.326880137125651
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,4,1,balanced,0.19953600565592447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,4,1,balanced,0.19993066787719727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,4,1,balanced,0.20542399088541666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,4,1,power_law_1.01,0.1387519955635071
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,4,1,balanced,0.2091253399848938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,4,1,balanced,0.24732800324757895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,4,1,balanced,0.2582826614379883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,4,power_law_1.2,0.3991424083709717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,4,1,balanced,0.2670666575431824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,4,1,balanced,0.2572373350461324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,4,1,power_law_1.01,0.7857088088989258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,4,power_law_1.2,0.4089344024658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,4,1,balanced,0.27188267310460407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,4,1,balanced,0.29783467451731366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,4,1,balanced,0.3163253267606099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,4,power_law_1.2,0.44884481430053713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,4,1,power_law_1.01,0.8116415977478028
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,4,1,balanced,0.45664532979329425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,4,1,balanced,0.4377439816792806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,4,1,power_law_1.01,0.9148608207702636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,4,1,balanced,0.9281067053476969
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,4,1,balanced,0.6811306476593018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,4,1,power_law_1.01,0.9960191726684571
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,4,1,balanced,0.9394559860229492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,4,1,power_law_1.01,1.1930944442749023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,4,1,balanced,1.2869866689046223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,4,1,balanced,0.6679946581522623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,4,1,power_law_1.01,0.029766398668289184
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,4,1,balanced,1.2684213320414226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,4,1,power_law_1.01,0.04351359903812409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,4,1,power_law_1.01,0.977177619934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,4,1,power_law_1.01,0.06238080263137817
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,4,1,balanced,1.8299412727355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,4,1,power_law_1.01,0.08303999900817871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,4,1,power_law_1.01,1.2049920082092285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,4,1,power_law_1.2,0.07258880138397217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,4,1,power_law_1.01,0.10618879795074462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,4,1,power_law_1.01,0.20067200660705567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,4,1,power_law_1.2,0.09265919923782348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,4,1,power_law_1.01,0.15703680515289306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,4,1,power_law_1.01,0.3018176078796387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,4,1,power_law_1.01,1.622425651550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,4,1,power_law_1.01,0.1659775972366333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,4,1,balanced,4.752858797709147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,4,1,power_law_1.01,0.3136575937271118
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,4,1,power_law_1.01,0.17486720085144042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,4,1,balanced,0.9992213249206543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,4,1,power_law_1.01,0.32563199996948244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,4,1,power_law_1.01,1.4243200302124024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,4,1,power_law_1.01,0.18444160223007203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,4,power_law_1.2,0.46555519104003906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,4,1,power_law_1.01,2.056287956237793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,4,1,power_law_1.01,0.33500161170959475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,4,1,power_law_1.01,0.18801920413970946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,4,power_law_1.2,0.4915071964263916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,4,1,power_law_1.01,0.3347264051437378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,4,1,power_law_1.01,1.8371456146240235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,4,power_law_1.2,0.5184127807617187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,4,power_law_1.2,0.5481599807739258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,4,1,power_law_1.01,2.9169279098510743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,4,1,power_law_1.2,0.14396159648895263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,4,1,balanced,6.1485544840494795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,4,power_law_1.2,0.624512004852295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,4,1,power_law_1.01,2.288319969177246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,4,1,power_law_1.2,0.06437119841575623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,4,1,power_law_1.2,0.08041599988937378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,4,1,balanced,1.1143360137939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,4,1,power_law_1.2,0.11571199893951416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,4,1,power_law_1.01,0.2028736114501953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,4,1,power_law_1.2,0.17025279998779297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,4,1,power_law_1.01,3.784262466430664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,4,1,balanced,1.5819573402404785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,4,1,power_law_1.01,0.3388479948043823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,4,1,power_law_1.2,0.21502718925476075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,4,1,power_law_1.01,3.231283187866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,4,1,power_law_1.01,0.37569279670715333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,4,1,power_law_1.2,0.3598655939102173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,4,1,balanced,2.0670560201009116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,4,1,power_law_1.01,0.3923712015151978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,4,1,power_law_1.2,0.3764672040939331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,4,1,power_law_1.2,0.2413503885269165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,4,1,power_law_1.01,0.4173696041107178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,4,1,power_law_1.2,0.39560320377349856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,4,1,power_law_1.2,0.31944320201873777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,4,1,power_law_1.01,0.46918401718139646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,4,1,balanced,2.7925281524658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,4,1,power_law_1.2,0.43045759201049805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,4,1,power_law_1.2,0.504422378540039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,4,1,power_law_1.01,0.5032000064849853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,4,1,power_law_1.2,0.44742398262023925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,4,1,power_law_1.01,4.138905715942383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,4,1,power_law_1.01,5.505126571655273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,4,1,power_law_1.2,0.5388480186462402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,4,1,power_law_1.2,0.48793601989746094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,4,1,power_law_1.01,0.21497600078582763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,4,1,power_law_1.2,0.5667327880859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,4,1,power_law_1.01,0.22926719188690187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,4,1,power_law_1.2,0.5310207843780518
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,4,1,power_law_1.01,0.25867519378662107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,4,1,power_law_1.2,0.5917759895324707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,4,1,power_law_1.2,0.4650432109832764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,4,1,power_law_1.01,0.29228160381317136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,4,1,power_law_1.2,0.6102911949157714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,4,1,power_law_1.2,0.4971968173980713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,4,1,power_law_1.01,0.24792959690093994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,4,power_law_1.2,0.7584767818450928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,4,1,power_law_1.2,0.5360000133514404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,4,1,power_law_1.01,0.2819519996643066
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,4,1,power_law_1.01,0.3041599988937378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,4,power_law_1.2,0.8566783905029297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,4,1,power_law_1.01,6.1815040588378904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,4,1,power_law_1.01,0.349401593208313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,4,1,power_law_1.01,7.2297218322753904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,4,1,balanced,3.743258794148763
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,4,1,power_law_1.01,0.36163198947906494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,4,1,power_law_1.01,0.45061120986938474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,4,1,power_law_1.01,0.4578944206237793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,4,1,power_law_1.2,0.6575744152069092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,4,1,power_law_1.01,0.5503935813903809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,4,1,balanced,5.428917566935222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,4,1,power_law_1.2,0.5494016170501709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,4,1,power_law_1.2,0.7035136222839355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,4,1,power_law_1.01,0.7461376190185547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,4,1,power_law_1.2,0.5861631870269776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,4,1,power_law_1.2,0.7452928066253662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,4,power_law_1.2,1.1299200057983398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,4,1,power_law_1.2,0.6969344139099121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,4,1,power_law_1.01,8.588275146484374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,4,1,power_law_1.2,0.7596479892730713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,4,1,power_law_1.2,0.8024255752563476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,4,power_law_1.2,1.3858752250671387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,4,1,power_law_1.2,0.05496960282325745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,4,1,balanced,7.119413375854492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,4,1,power_law_1.2,1.0077759742736816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,4,1,power_law_1.2,0.07283200025558471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,4,balanced,0.09130666653315227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,4,balanced,0.10708266496658325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,4,balanced,0.1622986694176992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,4,1,power_law_1.2,1.221132755279541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,4,balanced,0.3015146652857463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,4,balanced,0.5197653373082479
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,4,1,power_law_1.01,0.9340928077697754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,4,balanced,0.9569653669993082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,4,1,power_law_1.2,1.6544832229614257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,4,1,power_law_1.2,0.7654975891113281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,4,balanced,1.4082667032877605
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,4,1,power_law_1.01,1.3092479705810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,4,balanced,1.4117387135823567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,4,1,power_law_1.2,2.0850431442260744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,4,balanced,1.4149866104125977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,4,power_law_1.2,1.9574079513549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,4,1,power_law_1.01,1.8446720123291016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,4,balanced,1.4207679430643718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,4,balanced,0.0788373351097107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,4,balanced,0.09454933802286784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,4,balanced,0.1322933336098989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,4,balanced,0.20190932353337607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,4,1,power_law_1.2,2.9450752258300783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,4,balanced,0.33340267340342206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,4,power_law_1.2,2.6336767196655275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,4,balanced,0.6294666528701782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,4,balanced,0.9260533650716146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,4,1,power_law_1.2,0.7920576095581054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,4,1,power_law_1.2,0.029292801022529603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,4,balanced,0.9307573636372884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,4,1,power_law_1.2,0.04470399916172028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,4,1,power_law_1.2,0.09950079917907714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,4,balanced,0.9334879716237386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,4,1,power_law_1.2,0.06188160181045532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,4,1,power_law_1.2,0.13222399950027466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,4,balanced,0.9393813610076904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,4,1,power_law_1.2,0.07901440262794494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,4,power_law_1.2,3.7134334564208986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,4,1,power_law_1.2,0.1990399956703186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,4,balanced,1.1514933109283447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,4,1,power_law_1.2,0.10486400127410889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,4,balanced,1.681872049967448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,4,1,power_law_1.2,0.2934079885482788
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,4,1,power_law_1.2,0.15726079940795898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,4,1,power_law_1.2,0.3089215993881226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,4,balanced,1.6918400128682454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,4,1,power_law_1.2,0.31888000965118407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,4,balanced,1.7045013109842937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,4,1,power_law_1.2,0.825011157989502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,4,balanced,1.7235360145568848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,4,1,power_law_1.2,0.929043197631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,4,balanced,1.7429332733154297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,4,1,power_law_1.2,1.0033280372619628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,4,balanced,1.1604639689127605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,4,balanced,1.764037291208903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,4,1,power_law_1.2,0.165555202960968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,4,balanced,1.17575470606486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,4,1,power_law_1.2,1.2040960311889648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,4,1,power_law_1.01,0.6415232181549072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,4,balanced,1.19650133450826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,4,balanced,1.8014666239420574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,4,1,power_law_1.2,0.330297589302063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,4,1,power_law_1.01,0.806214427947998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,4,1,power_law_1.2,0.3349503993988037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,4,1,power_law_1.2,1.4159168243408202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,4,balanced,1.8748960494995117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,4,1,power_law_1.2,0.3511615991592407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,4,1,power_law_1.01,0.7289599895477294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,4,1,power_law_1.2,0.3932672023773193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,4,balanced,1.9481493631998699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,4,1,power_law_1.2,1.8338432312011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,4,1,power_law_1.2,0.4025087833404541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,4,1,power_law_1.2,0.4257215976715088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,4,balanced,2.089168071746826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,4,1,power_law_1.2,0.1752832055091858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,4,power_law_1.2,4.8689216613769535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,4,1,power_law_1.2,0.47658238410949705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,4,1,power_law_1.2,2.2949119567871095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,4,balanced,1.2208693027496338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,4,balanced,2.2305493354797363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,4,balanced,1.243717352549235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,4,balanced,2.545093377431234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,4,balanced,1.2818986574808757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,4,1,power_law_1.01,0.9070015907287597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,4,1,power_law_1.2,3.2223232269287108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,4,balanced,1.36406405766805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,4,1,power_law_1.2,3.82672004699707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,4,1,power_law_1.01,1.0847743988037108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,4,balanced,1.4578347206115723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,4,balanced,3.4399305979410806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,4,1,power_law_1.2,0.18584959506988524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,4,1,power_law_1.2,0.5295936107635498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,4,balanced,1.617578665415446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,4,1,power_law_1.01,1.452518367767334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,4,1,power_law_1.2,0.6569024085998535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,4,1,power_law_1.2,4.224620819091797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,4,balanced,1.80077330271403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,4,balanced,4.0736799240112305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,4,1,power_law_1.2,0.8152959823608399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,4,1,power_law_1.01,1.8311807632446289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,4,balanced,2.1569172541300454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,4,1,power_law_1.2,0.7383359909057617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,4,1,power_law_1.2,0.913811206817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,4,balanced,3.065237363179525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,4,1,power_law_1.01,2.551257514953613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,4,balanced,5.631216049194336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,4,1,power_law_1.2,1.0986368179321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,4,1,power_law_1.2,6.196569442749023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,4,balanced,3.7997706731160483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,4,1,power_law_1.2,1.469977569580078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,4,1,power_law_1.01,3.2886337280273437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,4,1,power_law_1.2,1.8343168258666993
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,4,1,power_law_1.2,0.1893952012062073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,4,balanced,5.0867306391398115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,4,balanced,8.691989262898764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,4,1,power_law_1.2,2.561574363708496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,4,balanced,7.10698127746582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,4,1,power_law_1.2,0.20563840866088867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,4,1,power_law_1.2,0.21870079040527343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,4,1,power_law_1.2,0.23500800132751465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,4,1,power_law_1.01,4.740691375732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,4,1,power_law_1.2,0.26781439781188965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,4,1,power_law_1.2,0.30053119659423827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,4,balanced,13.612778981526693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,4,1,power_law_1.2,0.2527424097061157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,4,1,power_law_1.2,0.2868096113204956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,4,balanced,9.877007802327475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,4,1,power_law_1.01,6.190816116333008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,4,power_law_1.01,0.15029120445251465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,4,power_law_1.01,0.3055488109588623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,4,balanced,0.08088533580303192
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,4,1,power_law_1.2,0.3064831972122192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,4,balanced,0.09230933586756389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,4,power_law_1.01,0.5350080013275147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,4,balanced,0.11971732974052429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,4,balanced,0.18854933977127075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,4,1,power_law_1.2,0.3568639993667603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,4,balanced,0.34966933727264404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,4,power_law_1.01,0.5018688201904297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,4,balanced,0.6543360153834025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,4,balanced,0.9640586376190186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,4,power_law_1.01,0.7026688098907471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,4,balanced,0.966421365737915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,4,balanced,0.9727573394775391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,4,power_law_1.01,0.9104448318481445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,4,balanced,0.9784213701883951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,4,1,power_law_1.2,3.286643218994141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,4,1,power_law_1.2,5.562854385375976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,4,balanced,0.9904959996541342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,4,power_law_1.01,0.9490495681762695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,4,balanced,0.9993600050608317
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,4,1,power_law_1.2,0.3947520017623901
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,4,balanced,1.0102026462554932
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,4,1,power_law_1.2,0.4570496082305908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,4,balanced,1.0283733208974202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,4,1,power_law_1.2,0.46282238960266114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,4,balanced,1.0497546990712483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,4,balanced,1.0688107013702393
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,4,1,power_law_1.2,0.561894416809082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,4,balanced,1.1049706935882568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,4,balanced,1.1743199825286865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,4,1,power_law_1.2,0.7526144027709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,4,balanced,1.2435680230458577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,4,balanced,1.3738932609558105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,4,power_law_1.2,6.822303771972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,4,balanced,1.5032426516215007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,4,1,power_law_1.2,7.295276641845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,4,balanced,1.8228106498718262
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,4,balanced,0.03134933362404505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,4,balanced,0.03956799954175949
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,4,balanced,0.06246933341026306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,4,balanced,2.4908960660298667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,4,1,power_law_1.2,0.9462335586547852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,4,balanced,3.0894559224446616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,4,power_law_1.01,1.2287808418273927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,4,balanced,0.09268266956011455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,4,balanced,0.15028267105420431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,4,balanced,0.2557599941889445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,4,power_law_1.01,1.1256959915161133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,4,balanced,0.36447465419769287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,4,balanced,0.3678400119145711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,4,balanced,0.37010665734608966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,4,1,power_law_1.2,8.086201477050782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,4,balanced,0.3725706736246745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,4,power_law_1.01,1.2311039924621583
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,4,balanced,0.3793066740036011
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,4,balanced,0.3874560197194417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,4,balanced,0.39529601732889813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,4,power_law_1.01,0.11742719411849975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,4,balanced,0.38955732186635333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,4,power_law_1.01,1.4290431976318358
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,4,balanced,0.3968106508255005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,4,balanced,4.080431938171387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,4,balanced,0.4063520034154256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,4,balanced,0.40563201904296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,4,power_law_1.01,1.426527976989746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,4,power_law_1.01,0.10776319503784179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,4,power_law_1.01,1.5621439933776855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,4,balanced,5.6874237060546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,4,power_law_1.01,1.432044792175293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,4,power_law_1.01,0.19779200553894044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,4,1,power_law_1.2,4.72740478515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,4,balanced,0.4291093349456787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,4,power_law_1.01,0.3292543888092041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,4,power_law_1.01,1.4756480216979981
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,4,balanced,0.496560017267863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,4,balanced,0.47119998931884766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,4,balanced,0.9009013175964355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,4,power_law_1.01,1.4900287628173827
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,4,balanced,0.5822133223215739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,4,balanced,7.781781514485677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,4,balanced,0.5866293509801229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,4,1,power_law_1.2,1.3279295921325684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,4,balanced,0.6853760083516439
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,4,balanced,0.9635199705759684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,4,power_law_1.01,0.334495997428894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,4,power_law_1.01,0.47614078521728515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,4,power_law_1.01,0.18120319843292237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,4,power_law_1.01,0.5973375797271728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,4,power_law_1.01,0.32663679122924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,4,power_law_1.01,1.659814453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,4,power_law_1.01,0.3334719896316528
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,4,balanced,1.1339200337727864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,4,balanced,1.7946079572041829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,4,power_law_1.2,0.15219199657440186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,4,power_law_1.01,0.6072319984436035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,4,power_law_1.2,0.2995903968811035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,4,power_law_1.01,0.4639552116394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,4,power_law_1.01,0.7113152027130127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,4,power_law_1.01,0.6339712142944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,4,power_law_1.01,0.6598015785217285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,4,1,power_law_1.2,1.9939071655273437
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,4,power_law_1.01,0.05777279734611511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,4,power_law_1.01,0.8793727874755859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,4,power_law_1.2,0.11612160205841064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,4,power_law_1.01,1.7835840225219726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,4,power_law_1.01,0.08512639999389648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,4,power_law_1.2,9.783622741699219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,4,power_law_1.01,0.7329919815063477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,4,power_law_1.01,0.8815423965454101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,4,power_law_1.2,0.19511040449142455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,4,power_law_1.01,0.14001280069351196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,4,power_law_1.01,0.7969535827636719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,4,power_law_1.2,0.24315519332885743
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,4,power_law_1.01,0.14287999868392945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,4,power_law_1.2,0.33320960998535154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,4,power_law_1.01,0.20019838809967042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,4,power_law_1.2,0.3957632064819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,4,power_law_1.2,0.4111487865447998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,4,power_law_1.01,0.2549567937850952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,4,power_law_1.2,0.5176447868347168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,4,power_law_1.2,0.10769280195236205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,4,power_law_1.2,0.5893887996673584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,4,power_law_1.2,0.18178559541702272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,4,power_law_1.01,1.848691177368164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,4,power_law_1.2,0.6844543933868408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,4,power_law_1.2,0.2612927913665771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,4,power_law_1.01,1.0132224082946777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,4,power_law_1.2,0.314847993850708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,4,power_law_1.01,2.086675262451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,4,power_law_1.01,0.9713791847229004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,4,power_law_1.2,0.4294911861419678
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,4,power_law_1.01,0.24982399940490724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,4,power_law_1.2,0.6011583805084229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,4,power_law_1.2,0.729529619216919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,4,power_law_1.01,0.3118655920028687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,4,power_law_1.01,0.8569279670715332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,4,power_law_1.2,0.6548607826232911
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,4,power_law_1.01,0.31969280242919923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,4,power_law_1.2,0.9517056465148925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,4,power_law_1.01,0.8415871620178222
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,4,power_law_1.01,0.3177727937698364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,4,power_law_1.2,0.8146816253662109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,4,power_law_1.2,0.8145471572875976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,4,power_law_1.01,0.8297344207763672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,4,power_law_1.2,0.9577343940734864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,4,power_law_1.01,0.3194688081741333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,4,power_law_1.2,0.6606016159057617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,4,power_law_1.2,0.8646400451660157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,4,power_law_1.01,0.8544768333435059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,4,power_law_1.01,1.0614784240722657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,4,power_law_1.01,2.3927871704101564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,4,power_law_1.2,1.0982015609741211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,4,power_law_1.2,0.7669824123382568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,4,power_law_1.01,0.88155517578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,4,power_law_1.01,1.0911744117736817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,4,power_law_1.2,1.1929408073425294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,4,power_law_1.2,0.9566335678100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,4,power_law_1.01,0.9029824256896972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,4,power_law_1.01,1.0657343864440918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,4,power_law_1.2,1.0364416122436524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,4,power_law_1.2,1.0214336395263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,4,power_law_1.01,1.1036160469055176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,4,power_law_1.01,0.341977596282959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,4,power_law_1.2,0.9479616165161133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,4,power_law_1.2,1.3747136116027832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,4,power_law_1.01,0.339084792137146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,4,power_law_1.01,1.207097625732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,4,power_law_1.2,1.06693115234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,4,power_law_1.01,0.32965118885040284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,4,power_law_1.2,1.471782398223877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,4,power_law_1.01,1.3438976287841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,4,power_law_1.01,0.3640064001083374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,4,power_law_1.2,1.0858752250671386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,4,power_law_1.01,0.9203007698059082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,4,power_law_1.01,0.34773759841918944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,4,power_law_1.2,1.4357184410095214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,4,power_law_1.01,1.493446445465088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,4,power_law_1.2,1.169222354888916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,4,power_law_1.01,0.40067200660705565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,4,power_law_1.2,0.8305472373962403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,4,power_law_1.01,0.3997375965118408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,4,power_law_1.01,1.7960639953613282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,4,power_law_1.01,0.43528318405151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,4,power_law_1.01,2.9784128189086916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,4,power_law_1.01,0.4764544010162354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,4,power_law_1.01,2.114828872680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,4,power_law_1.01,0.5855807781219482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,4,power_law_1.01,0.9598079681396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,4,power_law_1.01,0.6265920162200928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,4,power_law_1.2,1.5333375930786133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,4,power_law_1.01,3.5697921752929687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,4,power_law_1.01,1.114463996887207
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,4,power_law_1.01,0.6929408073425293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,4,power_law_1.2,0.7313407897949219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,4,power_law_1.2,1.5286911964416503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,4,power_law_1.01,1.2203328132629394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,4,power_law_1.2,0.7968255996704101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,4,power_law_1.2,1.4627840042114257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,4,power_law_1.2,0.9300671577453613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,4,power_law_1.01,4.9207103729248045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,4,power_law_1.01,2.676095962524414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,4,power_law_1.2,0.9335167884826661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,4,power_law_1.2,1.675116729736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,4,power_law_1.2,0.8759296417236329
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,4,power_law_1.01,0.8463616371154785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,4,power_law_1.2,1.8047807693481446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,4,power_law_1.01,3.263091278076172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,4,power_law_1.2,0.9459136009216309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,4,power_law_1.01,1.4504639625549316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,4,power_law_1.01,1.0556544303894042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,4,1,power_law_1.2,6.174879837036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,4,power_law_1.2,1.9242048263549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,4,power_law_1.2,1.012012767791748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,4,power_law_1.01,1.6852415084838868
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,4,power_law_1.01,1.4427583694458008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,4,power_law_1.2,1.1231040000915526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,4,power_law_1.01,4.604748916625977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,4,power_law_1.2,1.2226816177368165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,4,power_law_1.01,2.700748825073242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,4,power_law_1.2,1.466585636138916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,4,power_law_1.01,6.172057723999023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,4,power_law_1.2,1.7038080215454101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,4,power_law_1.01,5.580172729492188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,4,power_law_1.2,2.2779008865356447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,4,power_law_1.01,2.0939903259277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,4,power_law_1.2,2.216275215148926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,4,power_law_1.2,2.5264896392822265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,4,power_law_1.2,1.2161855697631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,4,power_law_1.01,2.6134336471557615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,4,power_law_1.2,1.3715904235839844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,4,power_law_1.2,2.73939208984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,4,power_law_1.01,8.496076965332032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,4,power_law_1.2,3.2388031005859377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,4,power_law_1.2,1.5260543823242188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,4,power_law_1.01,8.135935974121093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,4,power_law_1.01,3.568191909790039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,4,1,balanced,0.07851733267307281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,4,1,balanced,0.09669333696365356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,4,power_law_1.2,3.690348815917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,4,power_law_1.2,3.819308853149414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,4,power_law_1.2,0.05711359977722168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,4,power_law_1.2,0.08483200073242188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,4,power_law_1.01,4.571123123168945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,4,power_law_1.2,0.11132160425186158
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,4,power_law_1.2,0.13337600231170654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,4,power_law_1.2,4.817382431030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,4,1,balanced,0.14147733648618063
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,4,power_law_1.2,0.19058560132980346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,4,1,balanced,0.2885493238766988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,4,power_law_1.2,5.0864513397216795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,4,1,balanced,0.521232008934021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,4,power_law_1.01,11.556089782714844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,4,power_law_1.2,0.23532800674438475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,4,1,balanced,0.9580427010854086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,4,power_law_1.01,10.859519958496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,4,1,balanced,0.07081600030263265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,4,1,balanced,0.08676266670227051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,4,1,balanced,0.11526933312416077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,4,1,balanced,0.21068799495697021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,4,1,balanced,0.36467734972635907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,4,1,balanced,0.6900959809621176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,4,1,balanced,0.9803199768066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,4,power_law_1.2,1.8680831909179687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,4,power_law_1.2,6.975007629394531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,4,power_law_1.2,6.550387573242188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,4,power_law_1.2,2.1399423599243166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,4,power_law_1.2,0.2517568111419678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,4,1,balanced,0.06615466872851054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,4,1,balanced,0.08509866396586101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,4,power_law_1.01,6.275968170166015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,4,1,balanced,0.11753066380818684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,4,1,balanced,0.20222399632136026
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,4,power_law_1.2,0.33564798831939696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,4,1,balanced,0.3861279884974162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,4,1,balanced,0.7050666809082031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,4,power_law_1.2,2.8043264389038085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,4,power_law_1.2,0.3329535961151123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,4,1,balanced,0.9801706473032633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,4,1,balanced,1.0428640047709148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,4,1,balanced,0.9817919731140137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,4,power_law_1.2,0.3013056039810181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,4,1,balanced,0.9842987060546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,4,power_law_1.2,0.32746880054473876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,4,1,balanced,0.9904053211212158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,4,power_law_1.2,8.747187042236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,4,power_law_1.2,0.330784010887146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,4,power_law_1.2,9.146316528320312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,4,power_law_1.2,0.3189248085021973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,4,power_law_1.2,0.3262784004211426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,4,1,balanced,1.0475786526997883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,4,power_law_1.2,0.34067840576171876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,4,1,balanced,1.0521546999613445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,4,1,balanced,1.4021066029866536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,4,power_law_1.2,0.35829761028289797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,4,1,balanced,1.0562773545583088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,4,power_law_1.2,3.42476806640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,4,power_law_1.2,0.35944960117340086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,4,1,balanced,1.0630346934000652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,4,power_law_1.2,0.41096320152282717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,4,1,balanced,1.070314645767212
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,4,power_law_1.2,0.43355522155761717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,4,1,balanced,1.0774880250295003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,4,power_law_1.01,8.409503936767578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,4,power_law_1.2,0.4892735958099365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,4,1,balanced,1.0905653635660808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,4,1,power_law_1.01,0.07659519910812378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,4,1,balanced,1.1036907037099202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,4,power_law_1.2,4.705702209472657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,4,power_law_1.2,0.678438377380371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,4,1,power_law_1.01,0.09587200284004212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,4,1,balanced,1.1163360277811687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,4,1,balanced,1.4054187138875325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,4,1,power_law_1.01,0.1391487956047058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,4,1,balanced,1.1405226389567058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,4,power_law_1.2,12.624787139892579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,4,1,balanced,1.4057920773824055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,4,1,power_law_1.01,0.2830143928527832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,4,1,balanced,1.17740265528361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,4,1,balanced,1.4075093269348145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,4,1,power_law_1.01,0.06660479903221131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,4,1,balanced,1.2162346839904785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,4,1,power_law_1.01,0.45946240425109863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,4,1,power_law_1.01,0.08533120155334473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,4,1,balanced,1.4099733034769695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,4,1,balanced,1.2870453198750813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,4,1,power_law_1.01,0.6572095870971679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,4,1,power_law_1.01,0.1130560040473938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,4,1,balanced,1.4172372817993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,4,power_law_1.2,6.205145645141601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,4,1,balanced,1.3617013295491536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,4,1,balanced,0.994154691696167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,4,1,power_law_1.01,0.20033280849456786
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,4,1,balanced,0.03128000100453695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,4,power_law_1.2,0.6544000148773194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,4,1,balanced,0.04146666576464971
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,4,1,balanced,0.05997333427270254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,4,1,balanced,1.0000267028808594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,4,1,balanced,1.5457599957784016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,4,1,balanced,0.08880533774693807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,4,1,power_law_1.01,0.32442240715026854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,4,1,balanced,0.1444586714108785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,4,1,balanced,1.0077386697133381
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,4,1,balanced,0.2542133331298828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,4,1,balanced,0.36613333225250244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,4,1,power_law_1.01,0.4519296169281006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,4,1,balanced,2.2677332560221353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,4,1,power_law_1.01,0.5477824211120605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,4,1,balanced,2.5830186208089194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,4,1,power_law_1.01,0.6723840236663818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,4,1,balanced,1.425439993540446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,4,1,power_law_1.01,0.7616511821746826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,4,power_law_1.2,8.185651397705078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,4,1,power_law_1.01,0.7072896003723145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,4,power_law_1.2,0.7628672122955322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,4,1,balanced,3.520613352457682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,4,1,power_law_1.01,0.9461759567260742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,4,1,balanced,1.018346627553304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,4,1,balanced,0.3693653345108032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,4,1,power_law_1.01,0.7261119842529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,4,1,balanced,0.37138664722442627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,4,1,balanced,1.0293599764506023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,4,power_law_1.2,0.9804736137390136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,4,1,power_law_1.01,0.992249584197998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,4,1,balanced,1.0625279744466145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,4,1,power_law_1.01,0.770201587677002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,4,1,balanced,1.109071969985962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,4,1,power_law_1.01,1.029203224182129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,4,power_law_1.2,1.1603839874267579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,4,1,power_law_1.01,0.7906879901885986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,4,1,balanced,1.1690080165863037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,4,1,balanced,1.439450740814209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,4,1,power_law_1.01,1.088435173034668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,4,1,balanced,1.2794079780578613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,4,1,power_law_1.01,0.8447104454040527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,4,power_law_1.2,1.4953856468200684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,4,1,balanced,1.4186399777730305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,4,1,power_law_1.01,1.1012160301208496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,4,1,power_law_1.01,0.9180864334106446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,4,1,balanced,0.37390931447347003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,4,1,balanced,0.3803306818008423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,4,1,balanced,1.6817760467529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,4,1,balanced,0.3870933453241984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,4,1,power_law_1.01,0.9823936462402344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,4,1,power_law_1.01,1.1586239814758301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,4,1,balanced,4.788768132527669
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,4,power_law_1.2,2.993606376647949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,4,power_law_1.2,11.455321502685546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,4,1,power_law_1.01,1.061580753326416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,4,1,power_law_1.01,1.2326208114624024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,4,1,balanced,1.451365311940511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,4,1,power_law_1.01,0.9620287895202637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,4,1,power_law_1.01,1.2968832015991212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,4,1,power_law_1.01,1.062649631500244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,4,1,balanced,6.694208145141602
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,4,1,balanced,0.39290666580200195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,4,1,power_law_1.01,1.387161636352539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,4,1,balanced,0.4031253258387248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,4,1,power_law_1.01,1.1617471694946289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,4,1,balanced,2.381733258565267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,4,1,power_law_1.2,0.07662720084190369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,4,1,power_law_1.2,0.09563519954681396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,4,1,power_law_1.01,1.388742446899414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,4,1,power_law_1.01,1.3719807624816895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,4,1,power_law_1.2,0.13948160409927368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,4,1,balanced,2.9080588022867837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,4,1,balanced,1.4618666966756184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,4,1,power_law_1.2,0.28526721000671384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,4,1,power_law_1.01,1.5130880355834961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,4,1,power_law_1.01,1.5758655548095704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,4,1,balanced,1.5842240651448567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,4,1,power_law_1.2,0.4101696014404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,4,1,balanced,3.8678506215413413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,4,1,power_law_1.01,1.6198976516723633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,4,1,balanced,1.6275787353515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,4,1,balanced,0.41437868277231854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,4,1,power_law_1.01,1.9694080352783203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,4,1,balanced,0.42716264724731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,4,1,power_law_1.01,0.06618880033493042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,4,1,balanced,0.4758559862772624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,4,1,balanced,0.5177599986394247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,4,1,power_law_1.01,1.7902656555175782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,4,1,balanced,0.5124479929606119
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,4,1,balanced,0.57259734471639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,4,1,balanced,5.393701553344727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,4,1,balanced,0.602506677309672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,4,1,power_law_1.01,1.9788032531738282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,4,1,balanced,0.6937973499298096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,4,1,balanced,0.8488427003224691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,4,1,power_law_1.2,0.5838399887084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,4,1,balanced,0.9061493078867594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,4,1,balanced,1.6779093742370605
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,4,1,power_law_1.01,0.02881920039653778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,4,1,power_law_1.2,0.6766143798828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,4,1,power_law_1.01,0.08409600257873535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,4,1,power_law_1.01,0.03696640133857727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,4,1,balanced,1.7851840655008953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,4,1,balanced,1.7723360061645508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,4,1,balanced,7.370197296142578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,4,1,power_law_1.01,2.398700714111328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,4,1,power_law_1.2,0.9496000289916993
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,4,1,power_law_1.01,0.05702400207519531
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,4,1,balanced,1.3565492630004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,4,1,balanced,1.872597376505534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,4,1,power_law_1.2,0.06513919830322265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,4,1,power_law_1.01,0.08537600040435792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,4,1,power_law_1.2,0.9758975982666016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,4,1,balanced,2.063429355621338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,4,1,balanced,2.0937652587890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,4,1,power_law_1.01,0.1314303994178772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,4,1,power_law_1.01,3.211231994628906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,4,1,power_law_1.2,0.9947392463684082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,4,1,power_law_1.01,2.3924800872802736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,4,1,balanced,2.9373652140299478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,4,1,power_law_1.01,0.11601920127868652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,4,1,power_law_1.2,1.0768320083618164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,4,1,power_law_1.01,0.2001408100128174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,4,1,power_law_1.01,2.7927423477172852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,4,1,power_law_1.01,4.047782516479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,4,1,power_law_1.01,0.3511552095413208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,4,1,power_law_1.01,0.5047232151031494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,4,1,power_law_1.01,0.1833024024963379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,4,1,power_law_1.01,0.6126016139984131
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,4,1,power_law_1.01,0.22518401145935057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,4,1,power_law_1.01,0.7347968101501465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,4,1,power_law_1.01,5.735686492919922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,4,1,power_law_1.2,0.08413439989089966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,4,1,power_law_1.01,0.7819392204284668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,4,1,power_law_1.2,0.11093120574951172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,4,1,power_law_1.2,0.20024960041046141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,4,1,power_law_1.01,0.8090559959411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,4,1,power_law_1.01,3.631897735595703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,4,1,power_law_1.01,0.8165056228637695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,4,1,power_law_1.01,0.2742399930953979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,4,1,power_law_1.2,0.06666880249977111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,4,1,power_law_1.2,1.107795238494873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,4,1,power_law_1.2,0.08339840173721313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,4,1,power_law_1.01,0.28695039749145507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,4,1,power_law_1.2,0.11531519889831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,4,1,power_law_1.2,1.1539456367492675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,4,1,power_law_1.01,4.477676773071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,4,1,power_law_1.2,0.20024960041046141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,4,1,power_law_1.2,0.3049535989761353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,4,1,power_law_1.2,1.2628160476684571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,4,1,power_law_1.2,0.42018561363220214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,4,1,balanced,3.4492692947387695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,4,1,power_law_1.2,1.334886360168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,4,1,power_law_1.2,0.49480319023132324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,4,1,power_law_1.01,7.396940612792969
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,4,1,power_law_1.01,0.29395840167999265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,4,1,power_law_1.2,1.4318464279174805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,4,1,power_law_1.2,0.6567615985870361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,4,1,balanced,4.336597442626953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,4,1,power_law_1.01,0.30618240833282473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,4,1,power_law_1.2,0.3214143991470337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,4,1,power_law_1.2,0.6871679782867431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,4,1,power_law_1.01,0.31322879791259767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,4,1,power_law_1.2,0.4664576053619385
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,4,1,power_law_1.01,0.3215872049331665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,4,1,power_law_1.01,0.8475263595581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,4,1,power_law_1.2,0.5630911827087403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,4,1,power_law_1.01,0.8721023559570312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,4,1,balanced,6.364890416463216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,4,1,power_law_1.01,0.3459968090057373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,4,1,power_law_1.2,0.7226560115814209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,4,1,power_law_1.01,0.9080575942993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,4,1,power_law_1.01,6.529695892333985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,4,1,power_law_1.01,0.36791040897369387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,4,1,power_law_1.2,1.4552127838134765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,4,1,power_law_1.2,0.7507520198822022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,4,1,power_law_1.01,0.9297280311584473
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,4,1,power_law_1.01,0.38300158977508547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,4,1,power_law_1.2,0.7203775882720947
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,4,1,power_law_1.01,0.4334591865539551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,4,1,power_law_1.2,1.5285120010375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,4,1,power_law_1.2,0.7855487823486328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,4,1,power_law_1.01,0.9681920051574707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,4,1,power_law_1.2,0.7446976184844971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,4,1,power_law_1.2,0.8226688385009766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,4,1,power_law_1.2,1.6305791854858398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,4,1,power_law_1.2,0.8016511917114257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,4,1,power_law_1.2,0.8278335571289063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,4,1,balanced,10.143834431966146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,4,1,power_law_1.2,1.8335296630859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,4,1,power_law_1.01,8.27676773071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,4,1,power_law_1.2,0.8382719993591309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,4,balanced,0.037445334096749626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,4,balanced,0.03647999962170919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,4,balanced,0.045781334241231285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,4,1,power_law_1.2,0.8567296028137207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,4,balanced,0.07022400200366974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,4,1,power_law_1.01,0.5388095855712891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,4,1,power_law_1.2,2.0433984756469727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,4,1,power_law_1.01,0.9775615692138672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,4,1,power_law_1.2,0.8959808349609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,4,1,power_law_1.01,0.48699522018432617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,4,1,power_law_1.01,1.076921558380127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,4,1,power_law_1.01,0.6031424045562744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,4,1,power_law_1.2,2.4480192184448244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,4,1,power_law_1.01,1.1627327919006347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,4,balanced,0.06055466830730438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,4,1,power_law_1.2,0.9368960380554199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,4,balanced,0.07396266857783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,4,1,power_law_1.01,0.584332799911499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,4,balanced,0.10591466228167216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,4,1,power_law_1.2,0.932863998413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,4,1,power_law_1.2,0.997708797454834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,4,1,power_law_1.2,0.030687999725341798
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,4,1,power_law_1.01,0.7516543865203857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,4,1,power_law_1.2,2.869196891784668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,4,1,power_law_1.2,0.948748779296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,4,1,power_law_1.2,1.0907072067260741
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,4,1,power_law_1.2,0.03815680146217346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,4,1,power_law_1.01,0.7555327892303467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,4,1,power_law_1.2,1.0220607757568358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,4,1,power_law_1.2,0.9940223693847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,4,1,power_law_1.2,0.058534401655197146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,4,1,power_law_1.01,0.9867648124694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,4,1,power_law_1.2,1.0916671752929688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,4,1,power_law_1.2,0.08823040127754211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,4,balanced,0.18755733966827393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,4,1,power_law_1.2,3.709471893310547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,4,1,power_law_1.01,1.3434880256652832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,4,1,power_law_1.2,0.12298879623413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,4,1,power_law_1.2,1.1975808143615723
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,4,1,power_law_1.01,1.1299967765808105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,4,1,power_law_1.2,0.17116160392761232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,4,1,power_law_1.01,1.5235648155212402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,4,1,power_law_1.2,1.415782356262207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,4,balanced,0.10076799988746643
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,4,1,power_law_1.2,0.21395199298858641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,4,balanced,0.1430400013923645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,4,1,power_law_1.01,1.525388813018799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,4,1,power_law_1.01,1.8797632217407227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,4,1,power_law_1.2,1.6162496566772462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,4,1,power_law_1.2,4.759411239624024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,4,balanced,0.33366398016611737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,4,balanced,0.3348426818847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,4,balanced,0.33896533648173016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,4,1,power_law_1.01,2.202764892578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,4,balanced,0.3859519958496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,4,1,power_law_1.2,2.030054473876953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,4,balanced,0.38703465461730957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,4,1,power_law_1.01,2.2297151565551756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,4,balanced,0.39158932367960614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,4,balanced,0.0507893313964208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,4,balanced,0.056261335810025535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,4,balanced,0.06705600023269653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,4,balanced,0.09673066933949788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,4,balanced,0.14660799503326416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,4,balanced,0.2656746705373128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,4,1,power_law_1.2,2.453766441345215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,4,1,power_law_1.2,0.2662400007247925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,4,1,power_law_1.2,1.1190784454345704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,4,1,power_law_1.01,2.957663917541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,4,1,power_law_1.2,6.603411102294922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,4,1,power_law_1.2,0.28065919876098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,4,1,power_law_1.2,1.1841919898986817
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,4,1,power_law_1.2,0.2827264070510864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,4,1,power_law_1.2,3.279999923706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,4,balanced,0.39857598145802814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,4,1,power_law_1.2,1.3748671531677246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,4,1,power_law_1.2,0.29587199687957766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,4,balanced,0.4037439823150635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,4,1,power_law_1.01,3.702649688720703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,4,balanced,0.4153546492258708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,4,balanced,0.2331999937693278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,4,balanced,0.42793599764506024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,4,balanced,0.2368746598561605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,4,balanced,0.2690986593564351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,4,balanced,0.438970685005188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,4,balanced,0.27324267228444415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,4,1,power_law_1.2,1.5768768310546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,4,balanced,0.275546669960022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,4,balanced,0.27903467416763306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,4,balanced,0.2833706736564636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,4,balanced,0.28999465703964233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,4,1,power_law_1.2,4.119718551635742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,4,balanced,0.29573333263397217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,4,balanced,0.3075573245684306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,4,1,power_law_1.2,1.9199232101440429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,4,balanced,0.319760004679362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,4,balanced,0.32891732454299927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,4,1,power_law_1.2,8.54961929321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,4,balanced,0.3529493411382039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,4,1,power_law_1.01,5.129747009277343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,4,balanced,0.39931201934814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,4,balanced,0.4397653341293335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,4,1,power_law_1.2,0.308569598197937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,4,1,power_law_1.2,2.28799991607666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,4,balanced,0.5227306683858236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,4,balanced,0.4594240188598633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,4,balanced,0.6157919963200887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,4,1,power_law_1.2,0.32126080989837646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,4,balanced,0.5058453480402628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,4,balanced,0.54966934521993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,4,balanced,0.8773866494496664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,4,1,power_law_1.2,0.34878079891204833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,4,1,power_law_1.2,5.797395324707031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,4,balanced,1.055888017018636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,4,1,power_law_1.2,0.36827518939971926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,4,1,power_law_1.2,3.021798324584961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,4,balanced,1.500991980234782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,4,1,power_law_1.2,0.3913280010223389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,4,1,power_law_1.01,6.5538688659667965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,4,1,power_law_1.2,0.4416512012481689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,4,balanced,0.23914666970570883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,4,balanced,1.950826644897461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,4,balanced,0.2893120050430298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,4,power_law_1.01,0.0686784029006958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,4,balanced,0.29339200258255005
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,4,1,power_law_1.2,0.5541376113891602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,4,balanced,0.29713600873947144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,4,power_law_1.01,0.0998207986354828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,4,balanced,0.30321067571640015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,4,1,power_law_1.2,3.741331100463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,4,balanced,0.31014400720596313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,4,1,power_law_1.2,0.5004799842834473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,4,balanced,0.6373066504796346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,4,balanced,2.848965326944987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,4,balanced,0.7333226998647054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,4,1,power_law_1.2,0.63405442237854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,4,1,power_law_1.2,7.4691650390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,4,balanced,1.0181279977162678
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,4,1,power_law_1.2,0.6115839958190918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,4,balanced,1.1945066452026367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,4,power_law_1.01,0.05921919941902161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,4,balanced,3.75051212310791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,4,power_law_1.01,0.09415040016174317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,4,balanced,1.6776320139567058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,4,power_law_1.01,0.11203199625015259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,4,1,power_law_1.2,5.177664184570313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,4,balanced,0.02693866689999898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,4,balanced,0.02924266705910365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,4,power_law_1.01,0.06039680242538452
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,4,balanced,0.03201599915822347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,4,power_law_1.01,0.11654399633407593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,4,balanced,0.050186668833096824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,4,balanced,2.157365322113037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,4,balanced,0.06880533198515575
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,4,balanced,0.10434666275978088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,4,power_law_1.01,0.15480320453643798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,4,power_law_1.01,0.16076799631118774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,4,power_law_1.01,0.20061440467834474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,4,balanced,3.371861457824707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,4,power_law_1.01,0.21040639877319336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,4,1,power_law_1.2,0.7692736148834228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,4,power_law_1.01,0.28032000064849855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,4,power_law_1.01,0.27952640056610106
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,4,1,power_law_1.2,0.7775936126708984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,4,balanced,0.32103466987609863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,4,1,power_law_1.2,6.583507537841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,4,balanced,4.6609493891398115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,4,balanced,0.3327466646830241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,4,power_law_1.01,0.0931007981300354
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,4,balanced,0.10655466715494792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,4,balanced,0.3435680071512858
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,4,balanced,0.10938133796056111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,4,balanced,0.3683199882507324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,4,1,power_law_1.2,1.0120896339416503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,4,balanced,0.41910401980082196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,4,balanced,0.4631040096282959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,4,balanced,0.5613439877827963
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,4,1,power_law_1.2,1.1537343978881835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,4,balanced,0.659877339998881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,4,balanced,0.948527971903483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,4,balanced,1.1431573232014973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,4,1,power_law_1.2,1.5555328369140624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,4,power_law_1.01,0.27390079498291015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,4,balanced,1.6302293141682942
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,4,balanced,0.11108266313870747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,4,power_law_1.01,0.10364160537719727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,4,balanced,0.11341333389282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,4,balanced,0.11646933356920879
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,4,balanced,0.11219200491905212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,4,power_law_1.01,0.12086399793624877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,4,balanced,0.11527466773986816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,4,balanced,0.11897066235542297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,4,balanced,2.1212587356567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,4,balanced,0.11614933609962463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,4,power_law_1.01,0.1579327940940857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,4,1,power_law_1.2,2.353887939453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,4,power_law_1.01,0.029862400889396668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,4,power_law_1.01,0.20453760623931885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,4,power_law_1.2,0.0606719970703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,4,power_law_1.01,0.047539201378822324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,4,balanced,3.104703903198242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,4,power_law_1.2,0.09472000002861022
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,4,power_law_1.01,0.054681599140167236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,4,power_law_1.2,0.09052799940109253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,4,power_law_1.01,0.2916287899017334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,4,power_law_1.01,0.06503040194511414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,4,power_law_1.2,0.11678719520568848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,4,power_law_1.01,0.3032576084136963
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,4,power_law_1.01,0.0730239987373352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,4,power_law_1.2,0.1450816035270691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,4,power_law_1.01,0.3140160083770752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,4,power_law_1.01,0.07492480278015137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,4,balanced,0.11778133114178975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,4,power_law_1.2,0.17866239547729493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,4,balanced,0.12574932972590128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,4,power_law_1.01,0.327129602432251
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,4,balanced,0.1486026644706726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,4,power_law_1.01,0.09899520277976989
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,4,balanced,0.13843199610710144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,4,power_law_1.2,0.20548479557037352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,4,balanced,0.24703466892242432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,4,balanced,0.1734559933344523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,4,balanced,0.1844586730003357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,4,power_law_1.2,0.2122175931930542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,4,balanced,0.20631466309229532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,4,balanced,0.30161599318186444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,4,power_law_1.2,0.28102400302886965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,4,balanced,0.3510560194651286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,4,power_law_1.2,0.06747519969940186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,4,balanced,0.4983893235524495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,4,power_law_1.2,0.2818432092666626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,4,power_law_1.2,0.09902719855308532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,4,balanced,0.6445600191752116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,4,balanced,4.083717346191406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,4,power_law_1.2,0.27603199481964114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,4,power_law_1.2,0.11555839776992798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,4,power_law_1.2,0.2946880102157593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,4,power_law_1.2,0.17178879976272582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,4,power_law_1.01,0.3417855978012085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,4,power_law_1.2,0.18805760145187378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,4,power_law_1.01,0.10112639665603637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,4,power_law_1.01,0.24535679817199707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,4,power_law_1.01,0.3737087965011597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,4,power_law_1.2,0.22461440563201904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,4,power_law_1.01,0.09932799935340882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,4,power_law_1.01,0.23580799102783204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,4,power_law_1.01,0.44762239456176756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,4,power_law_1.2,0.27324159145355226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,4,power_law_1.01,0.10312960147857667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,4,power_law_1.01,0.2537087917327881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,4,power_law_1.01,0.5091008186340332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,4,power_law_1.2,0.29111039638519287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,4,power_law_1.01,0.25294079780578616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,4,power_law_1.2,0.3355776071548462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,4,power_law_1.01,0.2558336019515991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,4,power_law_1.2,0.35808000564575193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,4,power_law_1.2,0.06083199977874756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,4,power_law_1.01,0.2691967964172363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,4,power_law_1.2,0.37138559818267824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,4,power_law_1.01,0.13178240060806273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,4,power_law_1.2,0.09297279715538025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,4,power_law_1.01,0.2753920078277588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,4,power_law_1.2,0.3834304094314575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,4,power_law_1.2,0.09694719910621644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,4,power_law_1.01,0.18094079494476317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,4,power_law_1.01,0.2894399881362915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,4,power_law_1.2,0.02999039888381958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,4,power_law_1.2,0.3896192073822021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,4,power_law_1.2,0.11815040111541748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,4,power_law_1.01,0.19719680547714233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,4,power_law_1.01,0.3002687931060791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,4,power_law_1.2,0.044838398694992065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,4,power_law_1.2,0.41144962310791017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,4,power_law_1.2,0.1518399953842163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,4,power_law_1.01,0.22620799541473388
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,4,power_law_1.2,0.05115519762039185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,4,power_law_1.01,0.6277184009552002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,4,power_law_1.2,0.19527039527893067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,4,power_law_1.01,0.2735743999481201
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,4,power_law_1.2,0.061894398927688596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,4,power_law_1.01,0.7674560070037841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,4,power_law_1.2,0.3027967929840088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,4,power_law_1.01,0.283571195602417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,4,power_law_1.2,0.06780800223350525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,4,power_law_1.2,0.3108479976654053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,4,power_law_1.01,0.33580799102783204
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,4,power_law_1.2,0.07152000069618225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,4,power_law_1.01,1.0225536346435546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,4,power_law_1.2,0.3234944105148315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,4,power_law_1.01,0.3703552007675171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,4,power_law_1.2,0.33832321166992185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,4,power_law_1.01,1.3082304000854492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,4,power_law_1.2,0.3931967973709106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,4,power_law_1.01,0.31087360382080076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,4,power_law_1.2,0.4128896236419678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,4,power_law_1.01,0.3436736106872559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,4,power_law_1.2,0.43036799430847167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,4,power_law_1.2,0.24662399291992188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,4,power_law_1.01,0.40825600624084474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,4,power_law_1.2,0.4681920051574707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,4,power_law_1.2,0.23252480030059813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,4,power_law_1.01,0.460422420501709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,4,power_law_1.2,0.5400576114654541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,4,power_law_1.2,0.229420804977417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,4,power_law_1.2,0.09712640047073365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,4,power_law_1.01,0.5835519790649414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,4,power_law_1.2,0.24403839111328124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,4,power_law_1.2,0.09555839896202087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,4,power_law_1.2,0.25253119468688967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,4,power_law_1.01,0.7012864112854004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,4,power_law_1.2,0.10538239479064941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,4,power_law_1.2,0.2587199926376343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,4,power_law_1.2,0.4483071804046631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,4,power_law_1.01,0.9326975822448731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,4,power_law_1.2,0.2695103883743286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,4,power_law_1.2,0.5099071979522705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,4,power_law_1.2,0.2829632043838501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,4,power_law_1.2,0.29767680168151855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,4,power_law_1.2,0.6578368186950684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,4,power_law_1.2,0.3136127948760986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,4,power_law_1.2,0.5989183902740478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,4,power_law_1.2,0.3523519992828369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,4,power_law_1.2,0.7893504142761231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,4,power_law_1.2,0.4124800205230713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,4,power_law_1.2,0.7287168025970459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,4,power_law_1.2,1.0639616012573243
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,4,power_law_1.2,0.10323840379714966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,4,power_law_1.01,1.8411712646484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,4,power_law_1.2,0.8897791862487793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,4,power_law_1.2,0.10234880447387695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,4,power_law_1.2,1.3361984252929688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,4,power_law_1.2,0.10158720016479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,4,power_law_1.01,1.1793472290039062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,4,power_law_1.2,1.1309184074401855
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,4,power_law_1.2,0.10343680381774903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,4,power_law_1.01,2.463520050048828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,4,power_law_1.2,0.11111680269241334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,4,power_law_1.2,1.8979455947875976
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,4,power_law_1.2,0.10920959711074829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,4,power_law_1.01,1.6410367965698243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,4,power_law_1.2,0.4770048141479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,4,power_law_1.2,0.11114239692687988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,4,power_law_1.2,2.558572769165039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,4,power_law_1.01,3.4985408782958984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,4,power_law_1.01,2.149318313598633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,4,power_law_1.2,0.6064064025878906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,4,power_law_1.2,0.1201024055480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,4,power_law_1.2,0.13288960456848145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,4,power_law_1.2,0.709606409072876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,4,power_law_1.2,0.13982720375061036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,4,power_law_1.01,0.10895999670028686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,4,power_law_1.2,0.19408639669418334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,4,power_law_1.2,1.4093631744384765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,4,power_law_1.01,3.146668815612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,4,power_law_1.2,3.587788772583008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,4,power_law_1.01,4.59747200012207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,4,1,balanced,0.027386667827765148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,4,power_law_1.2,1.9708992004394532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,4,power_law_1.01,4.0634815216064455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,4,power_law_1.2,2.535980796813965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,4,power_law_1.01,0.37846400737762453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,4,1,balanced,0.02735999971628189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,4,power_law_1.2,0.9670592308044433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,4,1,balanced,0.04181866844495138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,4,1,balanced,0.05780800183614095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,4,1,balanced,0.09115200241406758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,4,power_law_1.01,0.10298240184783936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,4,1,balanced,0.14731733004252115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,4,1,balanced,0.24996266762415567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,4,power_law_1.01,0.1057088017463684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,4,1,balanced,0.2490826646486918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,4,1,balanced,0.25065600872039795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,4,power_law_1.01,0.11073280572891235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,4,power_law_1.01,0.10977920293807983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,4,power_law_1.2,3.7162303924560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,4,power_law_1.01,0.11113599538803101
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,4,power_law_1.01,0.12125439643859863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,4,power_law_1.01,0.13054720163345337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,4,power_law_1.2,4.7740222930908205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,4,power_law_1.01,0.13874560594558716
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,4,power_law_1.2,0.1934592008590698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,4,power_law_1.01,0.3870975971221924
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,4,power_law_1.01,0.17032320499420167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,4,power_law_1.2,0.23074560165405272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,4,power_law_1.01,0.39204480648040774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,4,power_law_1.01,0.19046399593353272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,4,1,balanced,0.046469335754712425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,4,power_law_1.2,4.91847038269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,4,power_law_1.01,0.40215039253234863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,4,power_law_1.01,0.22942719459533692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,4,power_law_1.01,0.40770559310913085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,4,power_law_1.01,0.26073598861694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,4,power_law_1.01,0.42421760559082033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,4,power_law_1.01,0.3355583906173706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,4,power_law_1.01,0.45729918479919435
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,4,power_law_1.01,0.4215551853179932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,4,power_law_1.01,0.5173120021820068
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,4,power_law_1.01,0.5757631778717041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,4,power_law_1.2,1.219328022003174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,4,power_law_1.01,0.5840832233428955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,4,1,balanced,0.2516746719678243
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,4,power_law_1.2,0.26368639469146726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,4,1,balanced,0.06428266565004985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,4,1,balanced,0.25205334027608234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,4,1,balanced,0.09872000416119893
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,4,power_law_1.01,1.0018752098083497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,4,power_law_1.01,0.7108928203582764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,4,power_law_1.2,0.34640638828277587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,4,power_law_1.2,1.7288000106811523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,4,1,power_law_1.01,0.02715519964694977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,4,power_law_1.2,0.4699264049530029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,4,1,power_law_1.01,0.04565759897232056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,4,1,balanced,0.02718399961789449
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,4,1,balanced,0.028581333657105763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,4,1,power_law_1.01,0.06104959845542908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,4,power_law_1.2,0.6606016159057617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,4,power_law_1.2,2.204550361633301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,4,1,power_law_1.01,0.09563519954681396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,4,1,power_law_1.01,0.15000319480895996
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,4,power_law_1.2,1.1255359649658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,4,1,power_law_1.01,0.18547840118408204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,4,1,power_law_1.01,0.027526399493217467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,4,1,balanced,0.1848213275273641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,4,1,balanced,0.34143467744191486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,4,1,power_law_1.01,0.25395200252532957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,4,1,power_law_1.01,0.033228799700737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,4,1,balanced,0.34282131989796955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,4,1,balanced,0.34065067768096924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,4,power_law_1.01,0.8327743530273437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,4,1,power_law_1.01,0.055276799201965335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,4,1,power_law_1.01,0.08880640268325805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,4,1,balanced,0.03160533308982849
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,4,1,balanced,0.04900266726811727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,4,power_law_1.01,1.1020799636840821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,4,1,power_law_1.01,0.1170240044593811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,4,1,power_law_1.01,0.14828799962997435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,4,power_law_1.01,1.3701120376586915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,4,1,power_law_1.01,0.19162240028381347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,4,1,balanced,0.2536906599998474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,4,1,power_law_1.01,0.19801599979400636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,4,power_law_1.2,3.36993293762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,4,1,power_law_1.01,0.26294400691986086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,4,1,balanced,0.34273068110148114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,4,1,balanced,0.34494932492574054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,4,1,power_law_1.01,0.26428799629211425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,4,1,balanced,0.3458079894383748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,4,1,balanced,0.3467093308766683
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,4,1,balanced,0.07077866792678833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,4,1,power_law_1.01,0.2681727886199951
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,4,1,balanced,0.10867200295130412
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,4,1,balanced,0.10973333319028218
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,4,1,balanced,0.1116373340288798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,4,1,power_law_1.01,0.27894399166107176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,4,1,balanced,0.1135093371073405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,4,1,balanced,0.11563199758529663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,4,1,balanced,0.1179146667321523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,4,1,balanced,0.12098133563995361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,4,power_law_1.2,4.189791870117188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,4,1,balanced,0.12387200196584065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,4,1,balanced,0.12776000301043192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,4,1,balanced,0.2568800052007039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,4,1,balanced,0.13945600390434265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,4,1,balanced,0.257914662361145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,4,1,balanced,0.14516799648602804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,4,1,balanced,0.2650773326555888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,4,1,balanced,0.15646933515866598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,4,power_law_1.01,1.9541248321533202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,4,1,balanced,0.1579146683216095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,4,1,balanced,0.03761066744724909
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,4,1,balanced,0.1696853240331014
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,4,1,balanced,0.18491733074188232
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,4,1,balanced,0.20286933581034342
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,4,1,balanced,0.2614133358001709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,4,1,balanced,0.3463253180185954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,4,1,balanced,0.2635040084520976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,4,1,balanced,0.35079999764760333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,4,1,balanced,0.5213013490041097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,4,1,power_law_1.01,0.035020801424980166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,4,1,balanced,0.3666079839070638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,4,1,balanced,0.41517333189646405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,4,1,balanced,0.37036800384521484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,4,1,power_law_1.01,0.2972223997116089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,4,1,balanced,0.3817760149637858
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,4,1,balanced,0.581712007522583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,4,1,balanced,0.40114665031433105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,4,1,power_law_1.01,0.31031041145324706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,4,1,balanced,0.7542986869812012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,4,1,power_law_1.01,0.3349760055541992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,4,1,power_law_1.01,0.20558080673217774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,4,1,balanced,0.2749706705411275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,4,1,balanced,0.2776266733805339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,4,1,power_law_1.01,0.3463615894317627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,4,1,power_law_1.01,0.2069632053375244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,4,1,balanced,0.2905226747194926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,4,1,balanced,0.31573333342870075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,4,1,power_law_1.01,0.3529983997344971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,4,1,balanced,0.33611734708150226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,4,1,balanced,0.38645867506663006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,4,1,power_law_1.01,0.37909760475158694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,4,power_law_1.01,2.515046310424805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,4,1,power_law_1.01,0.41515522003173827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,4,1,power_law_1.01,0.039705601334571836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,4,1,power_law_1.01,0.44590082168579104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,4,1,power_law_1.01,0.06044800281524658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,4,1,balanced,0.4246186812718709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,4,1,balanced,0.4680746793746948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,4,1,power_law_1.01,0.09848319888114929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,4,1,balanced,0.5215573310852051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,4,1,power_law_1.01,0.12810879945755005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,4,power_law_1.01,3.63135986328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,4,1,power_law_1.01,0.21946239471435547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,4,1,power_law_1.01,0.17596800327301027
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,4,1,power_law_1.01,0.02568320035934448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,4,1,power_law_1.01,0.23645439147949218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,4,1,power_law_1.01,0.21905920505523682
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,4,1,power_law_1.01,0.026649600267410277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,4,1,power_law_1.01,0.24636800289154054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,4,1,power_law_1.01,0.22592639923095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,4,1,power_law_1.01,0.03192960023880005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,4,1,power_law_1.01,0.2792639970779419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,4,1,power_law_1.01,0.23588480949401855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,4,1,power_law_1.01,0.25987839698791504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,4,1,power_law_1.01,0.24617600440979004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,4,power_law_1.01,4.649068832397461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,4,1,power_law_1.01,0.2666944026947021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,4,1,power_law_1.01,0.2526655912399292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,4,1,power_law_1.01,0.047628799080848695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,4,1,balanced,0.7049706776936849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,4,1,power_law_1.2,0.02712959945201874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,4,1,power_law_1.01,0.2589119911193848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,4,1,balanced,0.803050676981608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,4,1,power_law_1.2,0.046060800552368164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,4,1,power_law_1.01,0.2611840009689331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,4,1,power_law_1.2,0.06130560040473938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,4,1,balanced,1.1142240365346272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,4,1,balanced,0.4376586675643921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,4,1,power_law_1.01,0.26922879219055174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,4,1,power_law_1.2,0.09292160272598267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,4,1,balanced,0.6439786752065023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,4,1,balanced,1.4215413729349773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,4,1,power_law_1.01,0.2790080070495605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,4,1,power_law_1.2,0.13717119693756102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,4,1,balanced,0.7486826578776041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,4,1,power_law_1.01,0.2846528053283691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,4,1,power_law_1.2,0.18140159845352172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,4,1,balanced,1.0643359820048015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,4,1,power_law_1.01,0.5200319766998291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,4,1,power_law_1.01,0.3016576051712036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,4,1,power_law_1.2,0.2506239891052246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,4,1,power_law_1.01,0.34140160083770754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,4,1,power_law_1.01,0.28866560459136964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,4,1,power_law_1.2,0.261843204498291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,4,1,power_law_1.01,0.05950719714164734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,4,1,power_law_1.01,0.3812031984329224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,4,1,power_law_1.2,0.2629184007644653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,4,1,power_law_1.01,0.077183997631073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,4,1,power_law_1.2,0.2660736083984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,4,1,power_law_1.01,0.09218559861183166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,4,1,power_law_1.2,0.2911168098449707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,4,1,power_law_1.01,0.09662719964981079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,4,1,power_law_1.2,0.303769588470459
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,4,1,power_law_1.01,0.0992959976196289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,4,1,balanced,2.0600159962972007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,4,1,power_law_1.2,0.3118079900741577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,4,1,balanced,1.3815360069274902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,4,1,power_law_1.01,0.5807360172271728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,4,1,power_law_1.2,0.34955520629882814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,4,1,power_law_1.01,0.32963199615478517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,4,1,balanced,2.681856155395508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,4,1,power_law_1.2,0.347763204574585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,4,1,power_law_1.01,0.7321599960327149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,4,1,balanced,2.0023306210835776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,4,1,power_law_1.01,0.364851188659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,4,1,power_law_1.2,0.02837119996547699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,4,1,power_law_1.2,0.35320320129394533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,4,1,power_law_1.01,0.8875583648681641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,4,1,power_law_1.01,0.44218878746032714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,4,1,power_law_1.2,0.39312000274658204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,4,1,balanced,2.635807991027832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,4,1,power_law_1.2,0.4245567798614502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,4,1,power_law_1.01,1.2036800384521484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,4,1,power_law_1.2,0.4532991886138916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,4,1,power_law_1.01,0.101254403591156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,4,1,balanced,0.04572799801826477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,4,1,power_law_1.01,0.10554879903793335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,4,1,power_law_1.01,1.5134783744812013
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,4,1,power_law_1.01,0.11182080507278443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,4,1,power_law_1.01,0.4501823902130127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,4,1,power_law_1.01,0.11493760347366333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,4,1,power_law_1.2,0.03443840146064758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,4,1,power_law_1.01,0.5179967880249023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,4,1,power_law_1.01,0.12098560333251954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,4,1,power_law_1.2,0.03468799889087677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,4,1,power_law_1.2,0.05569919943809509
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,4,1,power_law_1.01,0.1310912013053894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,4,1,power_law_1.01,0.5151167869567871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,4,1,power_law_1.2,0.040524798631668094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,4,1,power_law_1.2,0.08756480216979981
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,4,1,power_law_1.01,0.138374400138855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,4,1,power_law_1.2,0.060127997398376466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,4,1,power_law_1.01,0.6711296081542969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,4,1,power_law_1.2,0.11096960306167603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,4,1,power_law_1.2,0.09649919867515563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,4,1,power_law_1.2,0.5266687870025635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,4,1,power_law_1.2,0.14179840087890624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,4,1,power_law_1.01,0.8180224418640136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,4,1,power_law_1.2,0.12094719409942627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,4,1,power_law_1.2,0.18756480216979982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,4,1,power_law_1.2,0.1675328016281128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,4,1,power_law_1.2,0.20142719745635987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,4,1,power_law_1.01,2.150643157958984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,4,1,power_law_1.2,0.2171072006225586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,4,1,power_law_1.2,0.2038655996322632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,4,1,power_law_1.01,0.6776000022888183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,4,1,power_law_1.2,0.20647039413452148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,4,1,power_law_1.2,0.22387840747833251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,4,1,power_law_1.2,0.22675840854644774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,4,1,power_law_1.01,0.15975040197372437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,4,1,power_law_1.01,0.8223103523254395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,4,1,power_law_1.2,0.23047680854797364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,4,1,power_law_1.2,0.24405760765075685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,4,1,power_law_1.01,0.15492479801177977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,4,1,power_law_1.01,2.792095947265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,4,1,power_law_1.2,0.25842559337615967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,4,1,power_law_1.01,1.125312042236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,4,1,power_law_1.2,0.5902719974517823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,4,1,power_law_1.2,0.28744959831237793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,4,1,power_law_1.01,1.1343040466308594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,4,1,power_law_1.2,0.26166400909423826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,4,1,power_law_1.2,0.750809621810913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,4,1,power_law_1.01,1.434227180480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,4,1,power_law_1.2,0.275161600112915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,4,1,power_law_1.01,1.44717435836792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,4,1,power_law_1.2,0.9054207801818848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,4,1,power_law_1.2,0.29687678813934326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,4,1,power_law_1.2,0.3406399965286255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,4,1,power_law_1.01,2.0328191757202148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,4,1,power_law_1.2,1.2189120292663573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,4,1,balanced,0.0617386649052302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,4,1,power_law_1.2,0.24062080383300782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,4,1,power_law_1.01,2.073747253417969
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,4,1,power_law_1.01,0.17812479734420777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,4,1,power_law_1.2,0.2532032012939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,4,1,power_law_1.01,0.18484480381011964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,4,1,power_law_1.2,1.5335807800292969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,4,1,power_law_1.2,0.25820159912109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,4,1,power_law_1.01,0.21780478954315186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,4,1,power_law_1.01,2.6335296630859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,4,1,power_law_1.2,0.26064000129699705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,4,balanced,0.03566399961709976
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,4,1,power_law_1.01,0.23467519283294677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,4,balanced,0.05907199780146281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,4,1,power_law_1.01,2.699212837219238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,4,balanced,0.07439466814200084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,4,1,power_law_1.2,0.2749567985534668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,4,balanced,0.10708266496658325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,4,balanced,0.035418666899204254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,4,1,power_law_1.01,0.2829056024551392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,4,1,power_law_1.2,0.2800512075424194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,4,1,power_law_1.01,0.3694015979766846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,4,1,power_law_1.2,0.29066879749298097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,4,1,balanced,0.10103999574979146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,4,1,power_law_1.01,0.45559039115905764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,4,1,power_law_1.2,0.31517438888549804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,4,1,power_law_1.2,0.3512831926345825
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,4,1,power_law_1.01,0.6269951820373535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,4,1,power_law_1.2,0.3916160106658936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,4,1,power_law_1.2,0.026470398902893065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,4,1,power_law_1.01,0.9156991958618164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,4,1,power_law_1.2,0.4599616050720215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,4,balanced,0.04404266675313314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,4,1,power_law_1.2,0.02687999904155731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,4,balanced,0.06841066479682922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,4,balanced,0.09714667002360027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,4,balanced,0.1421280006567637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,4,balanced,0.05043200155099233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,4,1,power_law_1.2,0.0338239997625351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,4,balanced,0.053717335065205894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,4,balanced,0.06736533343791962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,4,balanced,0.0960053304831187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,4,1,power_law_1.2,0.37448959350585936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,4,1,power_law_1.2,0.046598398685455324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,4,balanced,0.14594133694966635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,4,balanced,0.26703999439875287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,4,balanced,0.27085334062576294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,4,1,power_law_1.2,0.05858560204505921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,4,balanced,0.27078400055567425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,4,balanced,0.27526400486628216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,4,1,power_law_1.2,0.07404800057411194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,4,balanced,0.2776106595993042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,4,balanced,0.2835093339284261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,4,balanced,0.290063997109731
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,4,1,power_law_1.2,0.09140480160713196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,4,balanced,0.2938133279482524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,4,balanced,0.30792532364527386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,4,1,power_law_1.2,0.0966592013835907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,4,balanced,0.18929600715637207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,4,balanced,0.3182986577351888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,4,balanced,0.33429865042368573
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,4,1,power_law_1.2,0.09828479886054993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,4,balanced,0.33575467268625897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,4,balanced,0.2325119972229004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,4,balanced,0.33823998769124347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,4,balanced,0.23767467339833578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,4,1,power_law_1.2,0.10114560127258301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,4,balanced,0.38502931594848633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,4,balanced,0.23849600553512573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,4,balanced,0.290666659673055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,4,1,power_law_1.2,0.10611200332641602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,4,balanced,0.29205334186553955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,4,1,power_law_1.2,0.4495232105255127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,4,balanced,0.2964266737302144
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,4,1,power_law_1.2,0.10909440517425537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,4,1,power_law_1.2,0.5242496013641358
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,4,1,power_law_1.2,0.11507840156555176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,4,1,power_law_1.2,0.12163200378417968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,4,1,power_law_1.2,0.6810431957244873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,4,1,power_law_1.2,0.13198720216751098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,4,1,power_law_1.2,2.1722944259643553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,4,balanced,0.33025066057840985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,4,1,power_law_1.2,0.13918720483779906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,4,1,power_law_1.2,0.8383423805236816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,4,balanced,0.3524693250656128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,4,balanced,0.39843201637268066
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,4,1,power_law_1.2,0.16261759996414185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,4,balanced,0.4376266797383626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,4,balanced,0.3869493405024211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,4,1,power_law_1.2,1.1502911567687988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,4,balanced,0.5209013223648071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,4,balanced,0.3917440176010132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,4,balanced,0.3975040117899577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,4,balanced,0.30401599407196045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,4,balanced,0.40300798416137695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,4,balanced,0.30874133110046387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,4,balanced,0.4149706761042277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,4,balanced,0.3218986590703328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,4,1,power_law_1.2,1.4665087699890136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,4,balanced,0.4338933229446411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,4,balanced,0.33238933483759564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,4,balanced,0.43862934907277423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,4,balanced,0.4605546792348226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,4,balanced,0.5061866839726766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,4,1,power_law_1.2,0.539795207977295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,4,balanced,0.5491199890772501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,4,1,power_law_1.2,2.0981184005737306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,4,balanced,0.6362506548563639
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,4,1,power_law_1.2,0.15930240154266356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,4,balanced,0.7346293131510416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,4,1,power_law_1.2,0.182860803604126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,4,balanced,0.6165546576182047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,4,1,power_law_1.2,0.1876031994819641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,4,balanced,0.8729279836018881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,4,1,power_law_1.2,2.72475528717041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,4,1,power_law_1.2,0.22284159660339356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,4,1,balanced,0.1627253293991089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,4,balanced,1.0484480063120525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,4,balanced,0.34324800968170166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,4,1,balanced,0.2847040096918742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,4,balanced,0.36902932325998944
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,4,1,power_law_1.2,0.23879680633544922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,4,balanced,0.4172266721725464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,4,balanced,1.4936854044596355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,4,balanced,0.4620320002237956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,4,1,power_law_1.2,0.2881407976150513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,4,balanced,0.5631146828333536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,4,1,power_law_1.2,0.37486720085144043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,4,balanced,0.6598506768544515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,4,1,power_law_1.2,0.6902976036071777
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,4,1,power_law_1.2,0.46935038566589354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,4,balanced,0.948469320933024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,4,1,power_law_1.2,2.8019968032836915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,4,1,power_law_1.2,0.841158390045166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,4,power_law_1.01,0.06830080151557923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,4,1,power_law_1.2,0.6327616214752197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,4,power_law_1.01,0.09991040229797363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,4,1,power_law_1.2,1.139411163330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,4,1,balanced,0.28726933399836224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,4,power_law_1.01,0.13583999872207642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,4,1,balanced,0.2874880035718282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,4,1,balanced,0.2886506716410319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,4,power_law_1.01,0.17992960214614867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,4,1,balanced,0.2925386627515157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,4,1,power_law_1.2,1.4417535781860351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,4,1,balanced,0.29316800832748413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,4,power_law_1.01,0.1999295949935913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,4,1,balanced,0.2941439946492513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,4,1,balanced,0.29665066798528034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,4,1,balanced,0.30338666836420697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,4,power_law_1.01,0.2339263916015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,4,balanced,1.0187520186106365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,4,1,balanced,0.30710933605829877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,4,1,balanced,0.3112799922625224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,4,balanced,1.143999973932902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,4,balanced,1.1959253152211506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,4,balanced,1.6316159566243489
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,4,1,power_law_1.2,0.9720383644104004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,4,power_law_1.01,0.062003201246261595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,4,balanced,2.121077378590902
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,4,balanced,0.02666666607062022
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,4,balanced,0.027232001225153606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,4,balanced,1.942325274149577
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,4,balanced,0.03141333411137263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,4,balanced,0.048613334695498146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,4,balanced,0.0683840016523997
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,4,balanced,0.10392533739407857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,4,balanced,0.10538132985432942
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,4,balanced,0.10909866293271382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,4,balanced,3.1038506825764975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,4,power_law_1.01,0.2801471948623657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,4,1,balanced,0.32180267572402954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,4,balanced,2.8378400802612305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,4,1,balanced,0.3402080138524373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,4,power_law_1.01,0.26961278915405273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,4,1,balanced,0.35843201478322345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,4,balanced,1.679360071818034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,4,1,balanced,0.3920746644337972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,4,power_law_1.01,0.36110720634460447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,4,1,balanced,0.43884265422821045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,4,power_law_1.01,0.3630336046218872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,4,balanced,4.086629231770833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,4,balanced,3.732410748799642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,4,balanced,2.159850597381592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,4,power_law_1.01,0.09451519846916198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,4,power_law_1.01,0.10151679515838623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,4,power_law_1.01,0.06136959791183472
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,4,balanced,0.11029866337776184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,4,balanced,3.3909972508748374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,4,1,power_law_1.2,2.042521667480469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,4,balanced,0.11315199732780457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,4,balanced,0.11549866199493408
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,4,balanced,0.11307733257611592
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,4,balanced,0.11313600341478984
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,4,balanced,0.11753599842389424
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,4,balanced,0.11591466267903645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,4,balanced,0.1176533301671346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,4,balanced,0.12522666652997336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,4,balanced,0.14838932951291403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,4,1,balanced,0.6576106548309326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,4,balanced,0.1377173364162445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,4,balanced,0.24524267514546713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,4,balanced,4.655392011006673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,4,1,balanced,0.7362346649169922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,4,balanced,0.16865599155426025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,4,balanced,0.18503999710083008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,4,power_law_1.2,0.0686847984790802
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,4,balanced,0.2074986696243286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,4,1,balanced,1.0451467037200928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,4,balanced,0.30156266689300537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,4,power_law_1.2,0.09573760032653808
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,4,balanced,0.3513813416163127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,4,power_law_1.01,0.12451200485229492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,4,power_law_1.01,0.09304320216178893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,4,1,balanced,1.3618346850077312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,4,balanced,0.49859734376271564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,4,power_law_1.2,0.1401535987854004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,4,power_law_1.01,0.15516799688339233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,4,power_law_1.01,0.1023103952407837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,4,balanced,0.6452159881591797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,4,power_law_1.2,0.1640447974205017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,4,power_law_1.01,0.02967039942741394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,4,power_law_1.01,0.16458239555358886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,4,1,balanced,1.9812906583150227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,4,power_law_1.01,0.12833919525146484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,4,power_law_1.2,0.19445120096206664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,4,power_law_1.01,0.1908031940460205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,4,power_law_1.01,0.1639423966407776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,4,power_law_1.2,0.2230976104736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,4,power_law_1.01,0.19409279823303222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,4,power_law_1.01,0.3763904094696045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,4,power_law_1.01,0.18849920034408568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,4,1,power_law_1.2,2.6410688400268554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,4,power_law_1.2,0.25455360412597655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,4,power_law_1.01,0.2640320062637329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,4,power_law_1.01,0.38925440311431886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,4,power_law_1.01,0.23709440231323242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,4,power_law_1.01,0.2736063957214355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,4,power_law_1.2,0.29122560024261473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,4,power_law_1.01,0.38241279125213623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,4,power_law_1.01,0.24858880043029785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,4,power_law_1.01,0.2853760004043579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,4,power_law_1.2,0.06056320071220398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,4,power_law_1.2,0.354585599899292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,4,power_law_1.01,0.40259838104248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,4,power_law_1.01,0.2363840103149414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,4,power_law_1.01,0.28815999031066897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,4,power_law_1.2,0.09300479888916016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,4,power_law_1.2,0.3577023983001709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,4,power_law_1.01,0.4295167922973633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,4,power_law_1.01,0.24462718963623048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,4,power_law_1.2,0.05894399881362915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,4,power_law_1.01,0.30005760192871095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,4,power_law_1.2,0.09591680169105529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,4,power_law_1.01,0.4373311996459961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,4,power_law_1.01,0.25213439464569093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,4,power_law_1.2,0.08873599767684937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,4,power_law_1.01,0.3143104076385498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,4,power_law_1.2,0.1368127942085266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,4,power_law_1.01,0.4501503944396973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,4,power_law_1.01,0.32417280673980714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,4,1,balanced,2.61299737294515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,4,power_law_1.2,0.15827840566635132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,4,power_law_1.01,0.5046592235565186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,4,power_law_1.01,0.3371328115463257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,4,power_law_1.2,0.20547840595245362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,4,power_law_1.01,0.3654335975646973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,4,power_law_1.2,0.24444799423217772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,4,power_law_1.01,0.4437888145446777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,4,power_law_1.2,0.2306879997253418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,4,power_law_1.01,0.5022208213806152
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,4,power_law_1.01,0.04639999866485596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,4,power_law_1.2,0.3680255889892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,4,power_law_1.2,0.25978879928588866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,4,power_law_1.01,0.630022382736206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,4,power_law_1.2,0.09155840277671815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,4,power_law_1.01,0.05454080104827881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,4,power_law_1.2,0.24379520416259765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,4,power_law_1.2,0.11749119758605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,4,power_law_1.01,0.0639743983745575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,4,power_law_1.2,0.24204800128936768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,4,power_law_1.2,0.15212160348892212
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,4,power_law_1.01,0.06842880249023438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,4,power_law_1.2,0.2622528076171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,4,power_law_1.01,0.08069760203361512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,4,power_law_1.01,0.09510400295257568
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,4,power_law_1.01,0.10364160537719727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,4,power_law_1.2,0.029868799448013305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,4,power_law_1.01,0.2636800050735474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,4,power_law_1.01,0.09752320051193238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,4,power_law_1.2,0.04567039906978607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,4,power_law_1.01,0.10413440465927123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,4,power_law_1.01,0.7586751937866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,4,power_law_1.2,0.05065600275993347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,4,power_law_1.01,0.10572799444198608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,4,power_law_1.01,0.579257583618164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,4,power_law_1.2,0.060096001625061034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,4,power_law_1.01,1.0201600074768067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,4,power_law_1.01,0.10488959550857543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,4,power_law_1.2,0.15783040523529052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,4,power_law_1.2,0.07196159958839417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,4,power_law_1.01,0.10571520328521729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,4,power_law_1.2,0.18945280313491822
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,4,power_law_1.01,0.11102720499038696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,4,power_law_1.01,1.3094079971313477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,4,power_law_1.2,0.20064640045166016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,4,power_law_1.01,0.10729600191116333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,4,power_law_1.2,0.383078408241272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,4,power_law_1.01,1.8390655517578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,4,power_law_1.01,0.28023040294647217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,4,power_law_1.01,0.2864192008972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,4,power_law_1.01,0.2927295923233032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,4,power_law_1.2,0.25829761028289794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,4,power_law_1.01,0.6994751930236817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,4,power_law_1.01,0.3051647901535034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,4,power_law_1.2,0.292576003074646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,4,power_law_1.2,0.08047360181808472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,4,power_law_1.01,0.34530560970306395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,4,power_law_1.01,0.8503104209899902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,4,power_law_1.2,0.31448318958282473
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,4,power_law_1.2,0.09216639995574952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,4,power_law_1.01,0.40096001625061034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,4,power_law_1.2,0.25191679000854494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,4,power_law_1.2,0.31619839668273925
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,4,power_law_1.2,0.09824640154838563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,4,power_law_1.01,0.45518717765808103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,4,power_law_1.2,0.2730815887451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,4,power_law_1.2,0.3482048034667969
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,4,power_law_1.2,0.09807360172271729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,4,power_law_1.2,0.2873536109924316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,4,power_law_1.2,0.41198081970214845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,4,power_law_1.2,0.1011199951171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,4,power_law_1.2,0.28890879154205323
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,4,power_law_1.2,0.10623359680175781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,4,power_law_1.2,0.4732672214508057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,4,power_law_1.2,0.29564800262451174
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,4,power_law_1.2,0.10472320318222046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,4,power_law_1.2,0.30996479988098147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,4,power_law_1.2,0.5741759777069092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,4,power_law_1.2,0.10512640476226806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,4,power_law_1.2,0.328601598739624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,4,power_law_1.01,0.11067520380020142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,4,power_law_1.2,0.731443214416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,4,power_law_1.01,1.087820816040039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,4,power_law_1.01,0.11973119974136352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,4,power_law_1.2,0.389139199256897
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,4,power_law_1.01,0.1269055962562561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,4,power_law_1.01,0.584006404876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,4,power_law_1.01,1.3658944129943849
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,4,power_law_1.01,0.1372928023338318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,4,power_law_1.01,0.6934207916259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,4,power_law_1.01,0.20293118953704833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,4,power_law_1.01,2.369913673400879
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,4,power_law_1.01,0.19222400188446045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,4,power_law_1.01,0.9380415916442871
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,4,power_law_1.01,0.21634559631347655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,4,power_law_1.2,0.10878080129623413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,4,power_law_1.2,0.3420991897583008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,4,power_law_1.2,0.10792319774627686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,4,power_law_1.01,3.4621505737304688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,4,power_law_1.2,0.11269760131835938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,4,power_law_1.2,0.4036096096038818
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,4,power_law_1.2,0.12143360376358033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,4,power_law_1.2,0.4147712230682373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,4,power_law_1.2,0.13040000200271606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,4,power_law_1.2,0.4388607978820801
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,4,power_law_1.2,0.1444416046142578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,4,power_law_1.01,1.9047487258911133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,4,power_law_1.2,0.18645119667053223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,4,power_law_1.2,0.47624959945678713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,4,power_law_1.01,0.2719487905502319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,4,power_law_1.01,4.563404846191406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,4,power_law_1.2,0.19468799829483033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,4,1,balanced,0.027269333600997925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,4,power_law_1.2,0.5366144180297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,4,1,balanced,0.045519997676213585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,4,power_law_1.01,0.3548863887786865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,4,1,balanced,0.06205333272616068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,4,power_law_1.2,0.37318398952484133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,4,power_law_1.01,2.4290815353393556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,4,power_law_1.2,0.9784128189086914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,4,power_law_1.2,0.5943615913391114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,4,power_law_1.2,0.44834561347961427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,4,power_law_1.2,0.7211328029632569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,4,power_law_1.2,1.2066944122314454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,4,power_law_1.2,0.8678655624389648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,4,power_law_1.01,3.6223934173583983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,4,power_law_1.2,1.6863168716430663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,4,power_law_1.01,1.1836607933044434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,4,power_law_1.2,1.177670383453369
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,4,power_law_1.2,0.23006079196929932
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,4,power_law_1.01,0.4180287837982178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,4,power_law_1.2,0.2752959966659546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,4,power_law_1.2,1.4044672012329102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,4,power_law_1.01,1.6694400787353516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,4,power_law_1.2,2.2792320251464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,4,power_law_1.2,0.34715518951416013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,4,power_law_1.2,0.5190656185150146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,4,power_law_1.01,4.781689453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,4,power_law_1.2,0.661894416809082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,4,power_law_1.2,1.9837760925292969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,4,1,balanced,0.02943466603755951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,4,power_law_1.01,2.099807929992676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,4,power_law_1.2,0.7870912075042724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,4,power_law_1.2,3.3402305603027345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,4,1,balanced,0.09849599997202556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,4,1,balanced,0.18585066000620523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,4,power_law_1.2,1.065721607208252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,4,power_law_1.2,2.6366336822509764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,4,power_law_1.01,3.082374382019043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,4,power_law_1.01,0.5911808013916016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,4,power_law_1.2,1.3095808029174805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,4,power_law_1.2,0.44268798828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,4,power_law_1.01,1.0073792457580566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,4,1,balanced,0.03375466664632162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,4,1,balanced,0.05578133463859558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,4,power_law_1.2,4.295929718017578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,4,power_law_1.2,0.6021952152252197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,4,power_law_1.2,1.9428543090820312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,4,power_law_1.01,4.069670486450195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,4,power_law_1.2,1.1490943908691407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,4,1,power_law_1.01,0.026655998826026917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,4,1,power_law_1.01,0.045772799849510194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,4,1,balanced,0.03759466608365377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,4,1,balanced,0.04577599962552389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,4,1,balanced,0.060602664947509766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,4,1,power_law_1.01,0.061286401748657224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,4,1,balanced,0.10013332962989807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,4,1,balanced,0.16225066781044006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,4,1,power_law_1.01,0.0954367995262146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,4,1,balanced,0.09083732962608337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,4,1,power_law_1.01,0.15155199766159058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,4,power_law_1.2,3.695507049560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,4,1,power_law_1.01,0.18543360233306885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,4,1,power_law_1.01,0.25172479152679444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,4,1,balanced,0.3356693188349406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,4,1,balanced,0.33531733353932697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,4,1,power_law_1.01,0.26421120166778567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,4,1,balanced,0.3380906581878662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,4,power_law_1.2,2.586227226257324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,4,1,balanced,0.335968017578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,4,1,power_law_1.01,0.26545279026031493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,4,1,balanced,0.33804798126220703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,4,1,balanced,0.34426132837931317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,4,1,power_law_1.01,0.26933119297027586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,4,1,balanced,0.28549333413441974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,4,1,balanced,0.28668800989786786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,4,1,balanced,0.28733332951863605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,4,1,balanced,0.14814399679501852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,4,1,balanced,0.28751999139785767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,4,1,balanced,0.24992533524831137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,4,1,balanced,0.2918933431307475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,4,1,balanced,0.25059733788172406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,4,1,balanced,0.2940586606661479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,4,1,balanced,0.2511519988377889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,4,power_law_1.2,3.6856063842773437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,4,1,balanced,0.2523519992828369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,4,1,balanced,0.25439467032750446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,4,1,balanced,0.25411200523376465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,4,1,balanced,0.25787200530370075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,4,1,power_law_1.01,0.026943999528884887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,4,1,balanced,0.25805866718292236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,4,1,balanced,0.2648213307062785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,4,1,balanced,0.27425066630045575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,4,1,power_law_1.01,0.033036801218986514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,4,1,balanced,0.2781386574109395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,4,1,balanced,0.2910719911257426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,4,1,power_law_1.01,0.05485439896583557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,4,1,balanced,0.31484800577163696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,4,1,balanced,0.335968017578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,4,1,power_law_1.01,0.2822335958480835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,4,1,power_law_1.01,0.08942720293998718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,4,1,balanced,0.3848693370819092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,4,1,balanced,0.43860801060994464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,4,1,power_law_1.01,0.29937920570373533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,4,1,power_law_1.01,0.11913599967956542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,4,1,balanced,0.6414719820022583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,4,1,balanced,0.29439467191696167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,4,1,power_law_1.01,0.15230720043182372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,4,1,balanced,0.7466399669647217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,4,1,power_law_1.01,0.19258879423141478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,4,1,balanced,1.0603893597920735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,4,1,power_law_1.01,0.20004479885101317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,4,1,balanced,1.377562681833903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,4,1,balanced,0.34218132495880127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,4,1,power_law_1.01,0.20399999618530273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,4,1,balanced,0.34282131989796955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,4,1,balanced,0.34888001283009845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,4,power_law_1.2,5.15722885131836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,4,1,balanced,2.000197410583496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,4,1,balanced,0.369648019472758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,4,1,balanced,0.36981332302093506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,4,1,balanced,0.38076265652974445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,4,1,power_law_1.01,0.30259199142456056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,4,1,balanced,0.29499733448028564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,4,1,balanced,2.630565325419108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,4,1,balanced,0.3028480013211568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,4,1,power_law_1.01,0.3364032030105591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,4,1,balanced,0.30797332525253296
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,4,1,balanced,0.02548266698916753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,4,1,balanced,0.31249600648880005
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,4,1,balanced,0.027621333797772724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,4,1,power_law_1.01,0.34593920707702636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,4,1,balanced,0.3219839930534363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,4,1,balanced,0.338864008585612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,4,1,power_law_1.01,0.03532800078392029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,4,1,power_law_1.01,0.35304319858551025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,4,1,balanced,0.3577706813812256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,4,1,balanced,0.3925439914067586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,4,1,power_law_1.01,0.3702336072921753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,4,1,balanced,0.43825066089630127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,4,1,power_law_1.01,0.20643200874328613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,4,1,balanced,0.6580373446146647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,4,power_law_1.2,4.951225662231446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,4,1,balanced,0.7360853354136149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,4,1,balanced,0.40167466799418133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,4,1,power_law_1.2,0.026655998826026917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,4,1,balanced,0.42078399658203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,4,1,power_law_1.2,0.04625279903411865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,4,1,balanced,0.4665759801864624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,4,1,balanced,0.5211840073267618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,4,1,power_law_1.2,0.061452800035476686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,4,1,balanced,0.031685332457224526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,4,1,balanced,0.048138668139775596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,4,1,balanced,0.7072160243988037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,4,1,balanced,0.06957333286603291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,4,1,power_law_1.2,0.09329280257225037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,4,1,balanced,0.1076800028483073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,4,1,balanced,0.8018399874369303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,4,1,power_law_1.01,0.04017919898033142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,4,1,power_law_1.2,0.14302719831466676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,4,1,power_law_1.01,0.02705279886722565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,4,1,balanced,1.1127306620279949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,4,1,power_law_1.01,0.060761600732803345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,4,1,power_law_1.01,0.41991682052612306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,4,1,power_law_1.01,0.2191999912261963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,4,1,power_law_1.01,0.09841920137405395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,4,1,balanced,1.4223626454671223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,4,1,power_law_1.01,0.44377598762512205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,4,1,power_law_1.01,0.23605120182037354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,4,1,power_law_1.01,0.1383936047554016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,4,1,balanced,1.0395572980244954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,4,1,power_law_1.01,0.5112895965576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,4,1,power_law_1.01,0.24990720748901368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,4,1,power_law_1.01,0.17524479627609252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,4,1,balanced,2.0661813418070474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,4,1,balanced,1.3617919286092122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,4,1,power_law_1.01,0.5807807922363282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,4,1,power_law_1.01,0.2195199966430664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,4,1,power_law_1.01,0.732851219177246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,4,1,balanced,1.980778694152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,4,1,power_law_1.01,0.8827775955200196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,4,1,power_law_1.2,0.1825600028038025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,4,1,balanced,2.6133972803751626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,4,1,power_law_1.2,0.2545792102813721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,4,1,power_law_1.01,1.197753620147705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,4,1,power_law_1.2,0.2622720003128052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,4,1,power_law_1.2,0.26647679805755614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,4,1,power_law_1.01,0.2777728080749512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,4,1,power_law_1.01,1.510086441040039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,4,1,power_law_1.2,0.2719871997833252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,4,1,balanced,0.11107732852300008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,4,1,balanced,2.7012694676717124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,4,1,power_law_1.01,0.02736000120639801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,4,1,power_law_1.01,2.145055961608887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,4,1,power_law_1.01,0.03335039913654327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,4,1,power_law_1.2,0.027379199862480164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,4,1,power_law_1.2,0.03289600014686585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,4,1,power_law_1.01,2.775526428222656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,4,1,power_law_1.2,0.03518719971179962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,4,1,power_law_1.2,0.05498239994049072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,4,1,power_law_1.01,0.2604991912841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,4,1,power_law_1.01,0.23104000091552734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,4,1,power_law_1.2,0.042534399032592776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,4,1,balanced,0.1127359966437022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,4,1,power_law_1.01,0.2713279962539673
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,4,1,balanced,0.11274133125940959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,4,1,power_law_1.01,0.23538560867309571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,4,1,power_law_1.01,0.2888000011444092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,4,1,power_law_1.01,0.2469248056411743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,4,1,power_law_1.01,0.3266688108444214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,4,1,power_law_1.01,0.2518784046173096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,4,1,power_law_1.01,0.3667327880859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,4,1,power_law_1.01,0.0474368005990982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,4,1,power_law_1.01,0.25388801097869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,4,1,power_law_1.01,0.4412992000579834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,4,1,power_law_1.01,0.058348798751831056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,4,1,power_law_1.01,0.25994880199432374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,4,1,power_law_1.01,0.5171072006225585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,4,1,power_law_1.01,0.26856958866119385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,4,1,power_law_1.2,0.08768640160560608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,4,1,power_law_1.01,0.27479679584503175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,4,1,power_law_1.01,0.670739221572876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,4,1,balanced,0.1128480037053426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,4,1,power_law_1.2,0.10931199789047241
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,4,1,balanced,0.1181706686814626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,4,1,power_law_1.01,0.2813119888305664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,4,1,balanced,0.11928533514340718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,4,1,balanced,0.12218133608500163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,4,1,power_law_1.2,0.13878400325775148
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,4,1,balanced,0.1251626710096995
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,4,1,balanced,0.13663466771443686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,4,1,balanced,0.14200533429781595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,4,1,power_law_1.2,0.18457599878311157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,4,1,balanced,0.15272000432014465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,4,1,balanced,0.15627732872962952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,4,1,balanced,0.1701386570930481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,4,1,power_law_1.2,0.19441920518875122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,4,1,balanced,0.18331199884414673
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,4,1,balanced,0.20374933878580728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,4,1,power_law_1.2,0.20071039199829102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,4,1,balanced,0.2605973283449809
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,4,1,power_law_1.01,0.0787392020225525
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,4,1,balanced,0.2627786596616109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,4,1,power_law_1.2,0.20451838970184327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,4,1,power_law_1.2,0.29015679359436036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,4,1,balanced,0.5212533473968506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,4,1,power_law_1.01,0.09026560187339783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,4,1,power_law_1.2,0.060198402404785155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,4,1,power_law_1.2,0.22520959377288818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,4,1,power_law_1.2,0.3067008018493652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,4,1,power_law_1.01,0.09492480158805847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,4,1,power_law_1.2,0.09659519791603088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,4,1,power_law_1.2,0.24074881076812743
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,4,1,power_law_1.2,0.026713600754737853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,4,1,power_law_1.01,0.8249088287353515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,4,1,power_law_1.2,0.11825920343399048
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,4,1,power_law_1.2,0.027027198672294618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,4,1,power_law_1.2,0.168121600151062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,4,1,power_law_1.2,0.03351039886474609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,4,1,power_law_1.01,1.1372735977172852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,4,1,power_law_1.2,0.21866240501403808
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,4,1,power_law_1.2,0.04526079893112182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,4,1,power_law_1.2,0.22587521076202394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,4,1,power_law_1.01,1.4513343811035155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,4,1,power_law_1.2,0.23160960674285888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,4,1,balanced,0.4153013229370117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,4,1,power_law_1.2,0.24364159107208253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,4,1,power_law_1.2,0.31906559467315676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,4,1,power_law_1.01,0.30100479125976565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,4,1,power_law_1.2,0.25396480560302737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,4,1,power_law_1.01,2.0769279479980467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,4,1,power_law_1.2,0.3429759979248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,4,1,power_law_1.01,0.3391040086746216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,4,1,power_law_1.2,0.25816318988800047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,4,1,power_law_1.2,0.3461184024810791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,4,1,power_law_1.01,0.37946879863739014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,4,1,power_law_1.2,0.2598975896835327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,4,1,power_law_1.2,0.3510400056838989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,4,1,power_law_1.01,0.44875521659851075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,4,1,power_law_1.2,0.2691967964172363
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,4,1,power_law_1.2,0.055743998289108275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,4,1,power_law_1.01,0.5215295791625977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,4,1,power_law_1.2,0.07375360131263733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,4,1,power_law_1.01,0.6752448081970215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,4,1,balanced,0.5842719872792562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,4,1,power_law_1.2,0.08969600200653076
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,4,1,power_law_1.01,0.0976960003376007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,4,1,power_law_1.2,0.2571135997772217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,4,1,power_law_1.01,0.82107515335083
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,4,1,power_law_1.01,0.1014847993850708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,4,1,power_law_1.2,0.27933440208435056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,4,1,power_law_1.01,0.10585600137710571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,4,1,power_law_1.2,0.2616703987121582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,4,1,power_law_1.01,0.11096960306167603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,4,1,power_law_1.2,0.27436161041259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,4,1,power_law_1.2,0.39295361042022703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,4,1,power_law_1.2,0.2803071975708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,4,1,power_law_1.2,0.42465920448303224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,4,1,power_law_1.2,0.28663039207458496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,4,1,power_law_1.2,0.31529600620269777
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,4,1,power_law_1.2,0.09596160054206848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,4,1,power_law_1.2,0.35187840461730957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,4,1,power_law_1.2,0.09935359954833985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,4,1,power_law_1.2,0.3879231929779053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,4,1,power_law_1.01,1.1263936042785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,4,1,power_law_1.2,0.10165760517120362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,4,1,power_law_1.2,0.4614975929260254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,4,1,power_law_1.2,0.10584319829940796
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,4,1,power_law_1.01,0.11335680484771729
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,4,1,power_law_1.2,0.11020159721374512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,4,1,power_law_1.01,1.4251456260681152
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,4,1,power_law_1.01,0.12129919528961182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,4,1,power_law_1.2,0.11464320421218872
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,4,1,power_law_1.01,0.12921600341796874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,4,1,power_law_1.2,0.12165119647979736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,4,1,power_law_1.2,0.450816011428833
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,4,1,power_law_1.01,0.13827199935913087
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,4,1,power_law_1.2,0.13207679986953735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,4,1,balanced,0.7560426394144694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,4,1,power_law_1.01,2.0332799911499024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,4,1,power_law_1.2,0.5282559871673584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,4,1,power_law_1.2,0.1410752058029175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,4,balanced,0.03563733398914337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,4,balanced,0.049082666635513306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,4,1,power_law_1.2,0.1610303997993469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,4,1,power_law_1.2,0.5946176052093506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,4,1,power_law_1.2,0.1573632001876831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,4,1,power_law_1.2,0.7499904155731201
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,4,1,power_law_1.2,0.1806656002998352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,4,1,power_law_1.01,2.6329408645629884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,4,balanced,0.03569599986076355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,4,1,power_law_1.2,0.18804479837417604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,4,1,power_law_1.2,0.9081727981567382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,4,1,power_law_1.2,0.22136321067810058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,4,1,power_law_1.2,1.2190719604492188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,4,1,power_law_1.01,0.15781760215759277
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,4,1,power_law_1.01,0.1543295979499817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,4,balanced,0.06508799890677135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,4,balanced,0.0974826713403066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,4,balanced,0.16944533586502075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,4,balanced,0.2059146761894226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,4,balanced,0.2185759941736857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,4,balanced,0.22374399503072104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,4,1,power_law_1.2,0.29736320972442626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,4,balanced,0.22304532925287882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,4,balanced,0.22562134265899658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,4,balanced,0.03961066653331121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,4,1,power_law_1.2,0.33999359607696533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,4,1,power_law_1.2,0.37326080799102784
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,4,1,power_law_1.2,0.23739519119262695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,4,1,power_law_1.2,0.4496128082275391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,4,1,power_law_1.2,0.287558388710022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,4,1,power_law_1.2,0.5237696170806885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,4,1,power_law_1.2,0.37555201053619386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,4,1,power_law_1.2,0.6824512004852294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,4,1,power_law_1.2,0.464844799041748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,4,1,power_law_1.2,0.535916805267334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,4,balanced,0.2290560007095337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,4,balanced,0.05788266658782959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,4,balanced,0.23019200563430786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,4,balanced,0.09051199754079182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,4,balanced,0.23330666621526083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,4,balanced,0.24433600902557373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,4,balanced,0.25058666865030926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,4,balanced,0.256058673063914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,4,balanced,0.273087998231252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,4,balanced,0.3063093423843384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,4,balanced,0.3421706755956014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,4,balanced,0.48844265937805176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,4,1,power_law_1.2,1.5348928451538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,4,balanced,0.555733323097229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,4,1,power_law_1.01,2.7019327163696287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,4,balanced,0.7622559865315756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,4,balanced,0.9057439963022867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,4,power_law_1.01,0.06120960116386413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,4,1,power_law_1.2,0.6907199859619141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,4,1,power_law_1.2,2.172185516357422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,4,balanced,0.1165013313293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,4,balanced,0.14358400305112204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,4,balanced,1.3323040008544922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,4,1,power_law_1.2,0.8348480224609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,4,balanced,1.689242680867513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,4,1,power_law_1.01,0.17730560302734374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,4,1,power_law_1.2,2.8157184600830076
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,4,1,power_law_1.01,0.1830016016960144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,4,power_law_1.01,0.053478401899337766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,4,balanced,2.6051573753356934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,4,1,power_law_1.01,0.2182080030441284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,4,1,power_law_1.2,0.8328767776489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,4,1,power_law_1.2,0.6346687793731689
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,4,1,power_law_1.01,0.23374080657958984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,4,power_law_1.01,0.08124799728393554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,4,balanced,0.16479466358820596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,4,1,power_law_1.01,0.284006404876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,4,power_law_1.01,0.08617600202560424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,4,1,power_law_1.2,1.1496831893920898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,4,1,power_law_1.2,0.9285311698913574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,4,1,power_law_1.01,0.36805760860443115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,4,power_law_1.2,0.06164479851722717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,4,1,power_law_1.2,1.1346943855285645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,4,1,power_law_1.01,0.4595776081085205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,4,1,power_law_1.2,1.4607935905456544
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,4,1,power_law_1.01,0.6273727893829346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,4,power_law_1.01,0.0732800006866455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,4,1,power_law_1.2,1.4347071647644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,4,power_law_1.01,0.07982079982757569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,4,1,power_law_1.2,2.09051513671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,4,1,power_law_1.01,0.8878463745117188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,4,power_law_1.01,0.09446399807929992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,4,balanced,0.16858132680257162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,4,power_law_1.01,0.10179840326309204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,4,balanced,0.16945600509643555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,4,balanced,0.17075733343760172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,4,power_law_1.01,0.12179839611053467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,4,balanced,0.175818661848704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,4,balanced,0.17882666985193887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,4,power_law_1.01,0.12206720113754273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,4,power_law_1.2,0.07474560141563416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,4,1,power_law_1.2,2.7208511352539064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,4,power_law_1.01,0.1777791976928711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,4,1,balanced,0.026719999810059864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,4,power_law_1.2,0.09181439876556396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,4,1,balanced,0.039664000272750854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,4,1,balanced,0.056128000219662987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,4,power_law_1.01,0.2130431890487671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,4,1,balanced,0.08679999907811482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,4,power_law_1.2,0.10010240077972413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,4,1,balanced,0.1592639982700348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,4,1,balanced,0.1970240076382955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,4,power_law_1.2,0.1419648051261902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,4,1,balanced,0.19925866524378458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,4,1,balanced,0.1978399952252706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,4,1,balanced,0.19825067122777304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,4,1,balanced,0.1988746722539266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,4,1,balanced,0.20293333133061728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,4,balanced,3.640714645385742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,4,power_law_1.2,0.05537279844284058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,4,1,balanced,0.026170666019121807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,4,1,balanced,0.03030933439731598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,4,1,balanced,0.04794133206208547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,4,1,balanced,0.07699733475844066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,4,1,balanced,0.1176533301671346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,4,1,balanced,0.1486293375492096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,4,1,balanced,0.14870933691660562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,4,1,balanced,0.15031466881434122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,4,1,balanced,0.15180266896883646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,4,1,balanced,0.15033066272735596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,4,1,balanced,0.1520906686782837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,4,1,balanced,0.16059733430544534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,4,power_law_1.01,0.21247999668121337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,4,1,balanced,0.1609440048535665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,4,1,balanced,0.16776533921559653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,4,power_law_1.01,0.21263360977172852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,4,power_law_1.2,0.17322880029678345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,4,power_law_1.01,0.21313281059265138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,4,power_law_1.2,0.20421760082244872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,4,1,balanced,0.2076373298962911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,4,power_law_1.01,0.22038400173187256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,4,power_law_1.2,0.20241920948028563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,4,power_law_1.01,0.13127679824829103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,4,power_law_1.2,0.06778879761695862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,4,power_law_1.01,0.22581760883331298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,4,power_law_1.2,0.20890240669250487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,4,balanced,0.18551466862360635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,4,power_law_1.2,0.07816960215568543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,4,power_law_1.01,0.2283008098602295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,4,power_law_1.2,0.2160576105117798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,4,balanced,0.19502399365107217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,4,balanced,0.2021119991938273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,4,power_law_1.2,0.07935360074043274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,4,power_law_1.01,0.24833281040191652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,4,power_law_1.2,0.22455039024353027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,4,power_law_1.2,0.08923519849777221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,4,power_law_1.01,0.24934399127960205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,4,power_law_1.2,0.22437760829925538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,4,power_law_1.2,0.12028800249099732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,4,power_law_1.01,0.2771008014678955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,4,power_law_1.2,0.15070719718933107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,4,1,balanced,0.17094933986663818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,4,power_law_1.01,0.29297919273376466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,4,1,balanced,0.17509865760803223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,4,power_law_1.2,0.16047359704971315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,4,1,balanced,0.1865440011024475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,4,1,balanced,0.208079993724823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,4,power_law_1.2,0.16536320447921754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,4,1,balanced,0.23203200101852417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,4,1,balanced,0.21070400873819986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,4,1,balanced,0.3424533208211263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,4,1,balanced,0.215503990650177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,4,1,balanced,0.3806826670964559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,4,1,power_law_1.2,2.0332351684570313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,4,1,balanced,0.5327999989191691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,4,1,balanced,0.6236960093180338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,4,balanced,0.21262933810551962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,4,balanced,0.2312320073445638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,4,1,balanced,0.9293599923451742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,4,power_law_1.2,0.2294015884399414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,4,1,balanced,1.1779200236002605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,4,1,power_law_1.2,2.6436479568481444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,4,power_law_1.2,0.24615681171417236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,4,power_law_1.2,0.27414400577545167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,4,1,power_law_1.01,0.02492160052061081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,4,1,balanced,1.7315252621968586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,4,power_law_1.2,0.27686400413513185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,4,1,power_law_1.01,0.035667198896408084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,4,power_law_1.2,0.16569600105285645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,4,1,balanced,0.21622933944066366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,4,power_law_1.2,0.2853760004043579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,4,power_law_1.2,0.1702463984489441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,4,1,balanced,0.224506676197052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,4,1,balanced,2.286207993825277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,4,1,balanced,0.22826667626698813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,4,power_law_1.2,0.3392512083053589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,4,power_law_1.2,0.16970880031585694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,4,1,power_law_1.01,0.025017601251602174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,4,power_law_1.2,0.1811840057373047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,4,power_law_1.2,0.3914175987243652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,4,balanced,0.2719306747118632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,4,power_law_1.2,0.19601919651031494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,4,balanced,0.3081066608428955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,4,power_law_1.2,0.5091008186340332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,4,balanced,0.4402933518091838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,4,power_law_1.2,0.211244797706604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,4,balanced,0.5128213167190552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,4,power_law_1.2,0.6262400150299072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,4,power_law_1.2,0.22593278884887696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,4,balanced,0.721290667851766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,4,power_law_1.01,0.32743039131164553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,4,power_law_1.2,0.2439296007156372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,4,power_law_1.2,0.8804351806640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,4,power_law_1.01,0.3749056100845337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,4,power_law_1.2,0.30668160915374754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,4,1,power_law_1.01,0.054092800617218016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,4,power_law_1.01,0.15491199493408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,4,power_law_1.2,0.34895360469818115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,4,1,power_law_1.01,0.0795199990272522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,4,power_law_1.2,1.0898495674133302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,4,power_law_1.01,0.1628864049911499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,4,power_law_1.2,0.47312002182006835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,4,power_law_1.01,0.15762560367584227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,4,1,power_law_1.01,0.036371201276779175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,4,power_law_1.01,0.16513919830322266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,4,power_law_1.2,0.5859968185424804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,4,1,power_law_1.01,0.04859519898891449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,4,power_law_1.01,0.17074559926986693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,4,power_law_1.2,1.5828096389770507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,4,1,power_law_1.01,0.06909440159797668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,4,power_law_1.01,0.17479039430618287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,4,1,power_law_1.01,0.08922240138053894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,4,power_law_1.01,0.1792448043823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,4,1,power_law_1.01,0.11892479658126831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,4,power_law_1.01,0.19444479942321777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,4,power_law_1.2,1.9658111572265624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,4,1,power_law_1.01,0.12915199995040894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,4,power_law_1.01,0.20510079860687255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,4,1,power_law_1.01,0.13411200046539307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,4,power_law_1.01,0.21843841075897216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,4,1,balanced,0.2486506700515747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,4,1,power_law_1.01,0.13544319868087767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,4,1,balanced,0.26734399795532227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,4,1,balanced,0.3760693470637004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,4,1,power_law_1.01,0.14111360311508178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,4,1,balanced,0.4068640073140462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,4,1,power_law_1.01,0.1475775957107544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,4,1,balanced,0.5536319812138876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,4,1,power_law_1.01,0.14991359710693358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,4,1,balanced,0.6331679821014404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,4,power_law_1.2,0.8218688011169434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,4,1,power_law_1.01,0.1536064028739929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,4,1,balanced,0.9319413503011068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,4,1,power_law_1.01,0.16287360191345215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,4,balanced,0.8732906977335612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,4,power_law_1.2,1.0764863967895508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,4,1,balanced,1.1636266708374023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,4,balanced,1.2875733375549316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,4,power_law_1.01,0.4914048194885254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,4,1,balanced,1.7022773424784343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,4,power_law_1.2,1.4684608459472657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,4,power_law_1.01,0.5913728237152099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,4,balanced,1.652016003926595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,4,1,balanced,2.275866667429606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,4,power_law_1.01,0.8343296051025391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,4,1,power_law_1.2,0.025516799092292784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,4,balanced,2.4404640197753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,4,1,power_law_1.2,0.04017280042171478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,4,power_law_1.01,1.0394368171691895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,4,1,power_law_1.2,0.05358080267906189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,4,1,power_law_1.01,0.1722048044204712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,4,1,power_law_1.2,0.07652480006217957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,4,1,power_law_1.01,0.1782271981239319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,4,1,power_law_1.2,0.09344639778137206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,4,power_law_1.01,1.5222911834716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,4,1,power_law_1.01,0.20551679134368897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,4,1,power_law_1.2,0.15897599458694459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,4,power_law_1.01,0.24563839435577392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,4,1,power_law_1.01,0.236409592628479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,4,power_law_1.01,0.2938431978225708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,4,1,power_law_1.01,0.2632512092590332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,4,power_law_1.01,1.969593620300293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,4,power_law_1.01,0.340230393409729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,4,1,power_law_1.01,0.3312704086303711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,4,power_law_1.2,3.0163583755493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,4,power_law_1.01,0.45907201766967776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,4,1,power_law_1.01,0.3988480091094971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,4,power_law_1.01,0.5715007781982422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,4,1,power_law_1.01,0.5339392185211181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,4,power_law_1.01,2.8788991928100587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,4,balanced,3.2226292292277017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,4,1,power_law_1.01,0.6716224193572998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,4,1,power_law_1.2,0.027327999472618103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,4,1,power_law_1.01,0.10315519571304321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,4,1,power_law_1.2,0.16410239934921264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,4,1,power_law_1.2,0.037625598907470706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,4,1,power_law_1.01,0.16013439893722534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,4,1,power_law_1.2,0.1689919948577881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,4,1,power_law_1.01,0.16603519916534423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,4,1,power_law_1.2,0.17022720575332642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,4,1,power_law_1.01,0.1686784029006958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,4,1,power_law_1.2,0.17923200130462646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,4,1,power_law_1.01,0.16876159906387328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,4,power_law_1.01,3.4991870880126954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,4,1,power_law_1.2,0.1956480026245117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,4,1,power_law_1.01,0.17164160013198854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,4,balanced,0.1074186662832896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,4,1,power_law_1.2,0.20021119117736816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,4,1,power_law_1.01,0.19118080139160157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,4,1,power_law_1.01,0.197107195854187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,4,power_law_1.01,0.8057024002075195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,4,1,power_law_1.01,0.19898879528045654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,4,1,power_law_1.01,0.20921599864959717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,4,1,power_law_1.01,0.9521535873413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,4,power_law_1.01,1.0314623832702636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,4,1,power_law_1.01,0.21650559902191163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,4,1,power_law_1.2,0.04855040013790131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,4,1,power_law_1.01,0.23610239028930663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,4,1,power_law_1.2,0.0670144021511078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,4,1,power_law_1.01,0.25367040634155275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,4,1,power_law_1.01,1.2285632133483886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,4,1,power_law_1.2,0.07831680178642272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,4,power_law_1.01,1.4310848236083984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,4,1,power_law_1.01,0.2764287948608398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,4,1,power_law_1.2,0.1164736032485962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,4,power_law_1.2,1.9314048767089844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,4,1,power_law_1.01,0.2991679906845093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,4,balanced,0.18212266763051352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,4,balanced,0.31886933247248334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,4,1,power_law_1.01,1.787660789489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,4,1,power_law_1.2,0.20339200496673585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,4,power_law_1.01,1.8537343978881835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,4,balanced,0.5668960014979044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,4,1,power_law_1.2,0.21666560173034669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,4,balanced,1.0006720225016277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,4,1,power_law_1.2,0.22323200702667237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,4,1,power_law_1.2,0.24367361068725585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,4,power_law_1.2,2.8869056701660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,4,balanced,1.3878186543782551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,4,1,power_law_1.01,2.3411584854125977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,4,1,power_law_1.2,0.25525119304656985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,4,power_law_1.01,2.981772804260254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,4,balanced,1.3975680669148762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,4,balanced,0.08598400155703227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,4,balanced,0.13870933651924133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,4,balanced,0.21356799205144247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,4,balanced,1.404549280802409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,4,1,power_law_1.2,0.12917120456695558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,4,1,power_law_1.2,0.13680000305175782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,4,balanced,1.412768046061198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,4,1,power_law_1.2,0.1383296012878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,4,1,power_law_1.2,0.14625279903411864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,4,balanced,1.4246826171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,4,power_law_1.01,3.8330814361572267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,4,1,power_law_1.2,0.15682560205459595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,4,1,power_law_1.2,0.15345280170440673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,4,1,power_law_1.2,0.15685759782791137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,4,1,power_law_1.2,0.2802687883377075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,4,1,power_law_1.2,0.1698815941810608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,4,1,power_law_1.2,0.3085504055023193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,4,balanced,0.3669866720835368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,4,balanced,0.6628053188323975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,4,1,power_law_1.01,0.36683518886566163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,4,balanced,0.9306613604227701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,4,1,power_law_1.01,0.42223358154296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,4,balanced,0.9411893685658773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,4,balanced,0.08514666557312012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,4,1,power_law_1.01,0.5515711784362793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,4,balanced,0.13322133819262186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,4,balanced,0.9498506387074789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,4,balanced,0.20731200774510702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,4,balanced,1.4473387400309246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,4,balanced,0.3819253444671631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,4,balanced,0.9577279885609945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,4,balanced,0.7035893599192301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,4,balanced,0.9680480162302653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,4,balanced,1.769850730895996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,4,balanced,0.9897706508636475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,4,1,power_law_1.2,0.17725440263748168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,4,balanced,0.9880320231119791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,4,balanced,1.002570629119873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,4,1,power_law_1.2,0.18838399648666382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,4,balanced,1.320794661839803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,4,1,power_law_1.2,0.20881919860839843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,4,balanced,1.012437343597412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,4,power_law_1.2,4.589030456542969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,4,1,power_law_1.2,0.2383359909057617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,4,balanced,1.349199930826823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,4,balanced,1.0272586345672607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,4,1,power_law_1.2,0.27168641090393064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,4,balanced,1.0391039848327637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,4,1,power_law_1.2,0.3373055934906006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,4,power_law_1.01,0.2915071964263916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,4,1,power_law_1.01,0.6876031875610351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,4,balanced,1.065610647201538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,4,1,power_law_1.2,0.40293121337890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,4,1,power_law_1.01,0.9523648262023926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,4,1,power_law_1.2,0.5400896072387695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,4,1,power_law_1.2,0.3699392080307007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,4,balanced,1.7969120343526204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,4,1,power_law_1.2,0.6803135871887207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,4,1,power_law_1.01,1.2211199760437013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,4,power_law_1.2,3.960095977783203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,4,1,power_law_1.2,0.43333120346069337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,4,balanced,1.8399306933085124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,4,1,power_law_1.2,0.566054391860962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,4,balanced,1.3984266916910808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,4,balanced,1.8768107096354167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,4,balanced,1.4409119288126628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,4,balanced,1.9252586364746094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,4,power_law_1.01,0.511084794998169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,4,balanced,1.4993173281351726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,4,balanced,2.006293296813965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,4,balanced,1.6046667098999023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,4,1,power_law_1.2,0.9551936149597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,4,balanced,1.7976746559143066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,4,balanced,2.1777440706888833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,4,1,power_law_1.01,1.7630207061767578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,4,1,power_law_1.2,1.2378687858581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,4,1,power_law_1.2,0.6976128101348877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,4,balanced,1.9922240575154622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,4,balanced,2.363109270731608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,4,1,power_law_1.2,0.9687423706054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,4,1,power_law_1.01,2.310259246826172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,4,balanced,2.986069361368815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,4,1,power_law_1.2,1.2361663818359374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,4,power_law_1.01,0.5236480236053467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,4,balanced,3.2566452026367188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,4,power_law_1.01,0.7017151832580566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,4,power_law_1.01,0.1939136028289795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,4,balanced,3.35697078704834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,4,1,power_law_1.2,1.772153663635254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,4,balanced,3.6011041005452475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,4,power_law_1.01,0.3383807897567749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,4,power_law_1.01,0.9087167739868164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,4,balanced,1.087280035018921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,4,balanced,1.1125653584798176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,4,power_law_1.01,0.3505664110183716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,4,1,power_law_1.2,1.7991872787475587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,4,balanced,4.782426516215007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,4,balanced,1.1561066309611003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,4,power_law_1.01,1.3222784042358398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,4,power_law_1.01,0.48565120697021485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,4,balanced,1.1960533459981282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,4,power_law_1.01,0.19687680006027222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,4,power_law_1.01,1.1998208045959473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,4,power_law_1.01,0.3300287961959839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,4,balanced,5.642789204915364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,4,power_law_1.01,1.227020835876465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,4,1,power_law_1.2,2.3157440185546876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,4,power_law_1.01,1.246617603302002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,4,power_law_1.2,0.2909440040588379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,4,balanced,5.489973068237305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,4,balanced,1.2412799994150798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,4,balanced,1.3227413495381672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,4,balanced,9.364522933959961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,4,power_law_1.01,0.35099520683288576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,4,balanced,6.824431737263997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,4,power_law_1.01,0.4849408149719238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,4,power_law_1.01,0.6002304077148437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,4,power_law_1.01,0.6236735820770264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,4,power_law_1.01,0.7990848064422608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,4,power_law_1.01,0.8989567756652832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,4,balanced,1.4785173734029133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,4,power_law_1.01,0.8300671577453613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,4,balanced,1.6390773455301921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,4,power_law_1.01,0.8155072212219239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,4,balanced,12.426016489664713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,4,power_law_1.2,0.36750719547271726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,4,balanced,10.594298680623373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,4,power_law_1.01,0.8293439865112304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,4,power_law_1.2,0.524563217163086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,4,power_law_1.01,0.898521614074707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,4,power_law_1.2,0.658406400680542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,4,power_law_1.01,0.948793601989746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,4,power_law_1.01,1.015174388885498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,4,power_law_1.2,0.8020095825195312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,4,power_law_1.01,0.918284797668457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,4,power_law_1.01,1.2285375595092773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,4,power_law_1.01,0.9383232116699218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,4,power_law_1.2,1.2241408348083496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,4,power_law_1.01,1.3453375816345214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,4,power_law_1.01,0.915129566192627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,4,power_law_1.2,1.210431957244873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,4,power_law_1.01,1.5792575836181642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,4,balanced,13.850053151448568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,4,power_law_1.01,0.9537343978881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,4,1,power_law_1.2,2.355763244628906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,4,power_law_1.01,1.2141504287719727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,4,balanced,2.429471969604492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,4,power_law_1.01,1.6718591690063476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,4,power_law_1.01,0.9935935974121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,4,power_law_1.2,0.19514240026474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,4,power_law_1.01,1.2670080184936523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,4,balanced,19.13572820027669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,4,balanced,2.7501118977864585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,4,power_law_1.01,1.0343168258666993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,4,power_law_1.01,1.7219968795776368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,4,power_law_1.2,0.269651198387146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,4,power_law_1.01,1.361945629119873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,4,power_law_1.01,1.1188608169555665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,4,power_law_1.2,0.355020809173584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,4,power_law_1.01,1.7894271850585937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,4,balanced,3.9161974589029946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,4,power_law_1.01,1.4223872184753419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,4,power_law_1.2,1.2834688186645509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,4,power_law_1.01,1.1425536155700684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,4,power_law_1.2,0.47299838066101074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,4,power_law_1.01,1.8703680038452148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,4,power_law_1.01,1.5264703750610351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,4,power_law_1.2,0.5522496223449707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,4,power_law_1.2,1.2670207977294923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,4,power_law_1.01,1.252575969696045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,4,balanced,4.555573463439941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,4,power_law_1.2,0.8068608283996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,4,power_law_1.01,2.0113983154296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,4,power_law_1.01,1.360159969329834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,4,power_law_1.2,0.8005503654479981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,4,power_law_1.01,1.6199487686157226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,4,power_law_1.01,2.290176010131836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,4,power_law_1.2,0.8482815742492675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,4,balanced,20.92043685913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,4,power_law_1.2,0.8818367958068848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,4,power_law_1.01,1.9015552520751953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,4,balanced,6.890698750813802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,4,power_law_1.2,1.309075164794922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,4,power_law_1.01,2.6132415771484374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,4,power_law_1.2,0.911404800415039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,4,power_law_1.2,1.0133888244628906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,4,power_law_1.2,1.3918527603149413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,4,power_law_1.01,2.4703168869018555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,4,power_law_1.01,3.143929672241211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,4,power_law_1.2,1.2442560195922852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,4,power_law_1.2,1.6475967407226562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,4,balanced,24.958709716796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,4,power_law_1.2,1.3000384330749513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,4,power_law_1.01,3.0054527282714845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,4,power_law_1.01,1.6707263946533204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,4,power_law_1.2,1.6715263366699218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,4,power_law_1.2,1.3917247772216796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,4,power_law_1.01,3.815206527709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,4,power_law_1.01,2.004416084289551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,4,power_law_1.2,1.8032703399658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,4,power_law_1.2,1.475494384765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,4,power_law_1.01,4.086630249023438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,4,power_law_1.2,1.8428672790527343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,4,power_law_1.01,2.381113624572754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,4,balanced,8.69382413228353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,4,power_law_1.2,0.19596799612045288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,4,power_law_1.2,1.5553855895996094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,4,power_law_1.01,5.108332824707031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,4,power_law_1.2,0.2512063980102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,4,power_law_1.2,1.9333951950073243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,4,power_law_1.2,1.691142463684082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,4,power_law_1.01,2.993612861633301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,4,power_law_1.2,0.32193920612335203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,4,power_law_1.01,5.125388717651367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,4,power_law_1.2,2.0527103424072264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,4,power_law_1.2,0.464467191696167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,4,power_law_1.2,2.036582374572754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,4,power_law_1.2,0.5817408084869384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,4,power_law_1.2,2.4058752059936523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,4,power_law_1.01,6.417574310302735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,4,power_law_1.2,2.411065673828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,4,balanced,12.843018849690756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,4,power_law_1.2,0.929964828491211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,4,power_law_1.2,3.0320640563964845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,4,power_law_1.2,2.6458560943603517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,4,power_law_1.2,0.9142335891723633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,4,balanced,27.32238515218099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,4,1,balanced,0.09251733620961507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,4,1,balanced,0.16808533668518066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,4,power_law_1.01,7.428307342529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,4,power_law_1.2,0.9368191719055176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,4,power_law_1.2,3.6578048706054687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,4,power_law_1.2,3.3416255950927733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,4,power_law_1.2,0.9101696014404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,4,power_law_1.01,3.560870361328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,4,power_law_1.2,0.9529919624328613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,4,power_law_1.01,9.331027221679687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,4,power_law_1.2,0.9981951713562012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,4,1,balanced,0.2837173342704773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,4,power_law_1.2,4.01396484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,4,power_law_1.2,4.8924415588378904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,4,1,balanced,0.5172853469848633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,4,power_law_1.2,1.0261055946350097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,4,1,balanced,0.9622986316680908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,4,balanced,16.42950948079427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,4,1,balanced,1.401466687520345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,4,power_law_1.2,1.0545023918151855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,4,power_law_1.01,4.906028747558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,4,power_law_1.01,9.510688018798827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,4,1,balanced,0.08552533388137817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,4,1,balanced,1.4078933397928874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,4,1,balanced,0.12088533242543538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,4,1,balanced,0.19637332359949747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,4,power_law_1.2,1.1457280158996581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,4,1,balanced,1.4106240272521973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,4,power_law_1.2,5.403814315795898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,4,1,balanced,1.416202704111735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,4,power_law_1.2,1.2159359931945801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,4,power_law_1.2,6.6184638977050785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,4,1,balanced,1.4183947245279949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,4,1,balanced,1.4282399813334148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,4,power_law_1.2,1.2776512145996093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,4,power_law_1.01,6.207379150390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,4,1,balanced,1.4521759351094563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,4,power_law_1.01,11.98647689819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,4,1,balanced,0.36076800028483075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,4,1,balanced,1.4691413243611653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,4,1,balanced,0.6650400161743164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,4,power_law_1.2,1.436844825744629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,4,1,balanced,0.9638506571451823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,4,1,balanced,1.4776053428649902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,4,power_law_1.2,6.564358520507812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,4,1,balanced,0.9683039983113607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,4,1,balanced,1.4871145884195964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,4,1,balanced,0.9704053401947021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,4,power_law_1.2,1.692518424987793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,4,1,balanced,0.9738453229268392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,4,1,balanced,1.9477759997049968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,4,1,balanced,0.9767839908599854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,4,power_law_1.2,9.211103820800782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,4,1,balanced,0.9848106702168783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,4,power_law_1.2,1.9528831481933593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,4,power_law_1.01,14.006597900390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,4,1,balanced,0.9939733346303304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,4,power_law_1.01,8.875526428222656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,4,1,balanced,1.0064266522725422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,4,1,balanced,1.0202133655548096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,4,power_law_1.2,2.5189247131347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,4,1,balanced,1.0372533003489177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,4,1,balanced,1.4738880793253581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,4,power_law_1.2,10.136793518066407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,4,1,balanced,2.4898932774861655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,4,1,balanced,1.9275573094685872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,4,power_law_1.2,3.0709440231323244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,4,1,balanced,1.7042773564656575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,4,1,balanced,1.3446720441182454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,4,1,balanced,1.4580906232198079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,4,1,balanced,1.8058560689290364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,4,power_law_1.2,11.445574188232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,4,1,balanced,2.283397356669108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,4,power_law_1.01,18.26044158935547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,4,power_law_1.2,4.2740222930908205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,4,1,balanced,2.5885440508524575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,4,power_law_1.01,11.4487548828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,4,1,balanced,2.493504047393799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,4,1,balanced,2.816517194112142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,4,1,balanced,3.566234588623047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,4,1,balanced,4.08789857228597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,4,power_law_1.2,5.305574417114258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,4,power_law_1.01,17.868544006347655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,4,1,balanced,4.018074671427409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,4,1,power_law_1.01,0.0877568006515503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,4,power_law_1.2,13.269970703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,4,1,balanced,4.2583573659261065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,4,1,power_law_1.01,0.1611456036567688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,4,1,power_law_1.01,0.27903358936309813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,4,1,balanced,6.122287750244141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,4,1,power_law_1.01,0.4565375804901123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,4,power_law_1.2,7.794445037841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,4,1,balanced,6.690810521443685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,4,1,power_law_1.01,0.6624959945678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,4,power_law_1.01,16.51905212402344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,4,1,power_law_1.01,1.046668815612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,4,1,balanced,7.61082140604655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,4,1,power_law_1.01,1.1407039642333985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,4,1,balanced,8.42624537150065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,4,1,power_law_1.01,1.1647616386413575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,4,power_law_1.2,19.052120971679688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,4,1,power_law_1.01,1.1958016395568847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,4,power_law_1.01,24.3610107421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,4,power_law_1.2,9.623020935058594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,4,1,power_law_1.01,1.2216447830200194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,4,1,power_law_1.01,0.07273600101470948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,4,1,power_law_1.01,1.3024831771850587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,4,1,balanced,11.2673708597819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,4,1,power_law_1.01,0.11938560009002686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,4,power_law_1.2,20.40052490234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,4,1,power_law_1.01,0.19349759817123413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,4,1,power_law_1.01,1.3726592063903809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,4,1,power_law_1.01,0.3211456060409546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,4,1,balanced,12.942068735758463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,4,1,power_law_1.01,0.4544064044952393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,4,1,power_law_1.01,0.7309951782226562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,4,1,power_law_1.01,0.7773952007293701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,4,1,power_law_1.01,1.4683008193969727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,4,1,balanced,14.351509094238281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,4,power_law_1.2,14.338739013671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,4,1,power_law_1.01,1.6564863204956055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,4,1,power_law_1.2,0.08602880239486695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,4,power_law_1.01,23.975616455078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,4,1,power_law_1.2,0.16135679483413695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,4,1,power_law_1.01,1.8437759399414062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,4,1,power_law_1.2,0.07285119891166687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,4,1,power_law_1.2,0.2785151958465576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,4,1,power_law_1.01,0.8092288017272949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,4,1,balanced,16.71231969197591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,4,1,power_law_1.01,2.0452224731445314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,4,1,power_law_1.2,0.4329279899597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,8,balanced,0.17466133832931519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,4,1,power_law_1.01,0.8292415618896485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,8,balanced,0.29477866490681964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,8,balanced,0.30533333619435626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,8,balanced,0.3058240016301473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,4,1,power_law_1.2,0.5725247859954834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,4,1,power_law_1.01,0.8506303787231445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,4,1,power_law_1.01,2.4650367736816405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,4,1,power_law_1.01,0.9456768035888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,4,1,power_law_1.01,1.7489152908325196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,4,1,power_law_1.2,0.11834239959716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,4,1,power_law_1.01,1.0346624374389648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,4,1,power_law_1.2,0.1938048005104065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,4,1,power_law_1.01,1.1015040397644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,4,power_law_1.2,23.800901794433592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,4,1,power_law_1.01,1.9377471923828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,8,balanced,0.3033439914385478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,4,1,power_law_1.2,0.29569919109344484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,8,balanced,0.14452266693115234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,8,balanced,0.22762133677800497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,4,1,power_law_1.01,1.2634943962097167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,4,1,power_law_1.2,1.046784019470215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,4,1,power_law_1.2,0.4081600189208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,4,1,power_law_1.01,2.378124809265137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,4,power_law_1.2,27.951776123046876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,4,1,power_law_1.01,1.4124223709106445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,4,1,power_law_1.2,0.7109632015228271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,4,power_law_1.2,18.437350463867187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,4,1,power_law_1.01,1.5805439949035645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,4,1,power_law_1.2,0.7735616207122803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,4,1,power_law_1.01,2.7757823944091795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,8,balanced,0.296944002310435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,8,balanced,0.29711467027664185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,4,1,power_law_1.2,0.806771183013916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,8,balanced,0.2433919906616211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,8,balanced,0.3006719946861267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,8,balanced,0.237664004166921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,8,balanced,0.3073546687761943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,8,balanced,0.23722134033838907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,8,balanced,0.30298133691151935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,4,1,power_law_1.2,0.8446144104003906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,4,1,power_law_1.2,1.1301888465881347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,4,1,power_law_1.2,0.881868839263916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,4,1,power_law_1.01,3.695391845703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,4,1,power_law_1.2,0.970751953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,4,1,power_law_1.2,1.1706624031066895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,4,1,power_law_1.2,1.0688896179199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,4,1,power_law_1.2,1.1996671676635742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,4,1,power_law_1.01,1.9004352569580079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,8,balanced,0.24147733052571616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,8,balanced,0.24384532372156778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,8,balanced,0.2509920001029968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,4,1,power_law_1.01,4.618560028076172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,4,1,power_law_1.2,1.2663488388061523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,4,1,power_law_1.01,1.4832575798034668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,4,1,power_law_1.2,1.3400128364562989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,4,1,power_law_1.01,1.7030015945434571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,4,1,power_law_1.2,1.4070655822753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,4,1,power_law_1.01,2.1196544647216795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,4,1,power_law_1.2,1.1445568084716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,4,1,power_law_1.2,1.5354368209838867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,4,1,power_law_1.01,6.752339172363281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,4,1,power_law_1.2,1.3103808403015136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,4,1,power_law_1.2,1.7292352676391602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,4,1,power_law_1.2,1.4628800392150878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,4,1,power_law_1.2,1.9198591232299804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,8,balanced,0.31330666939417523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,8,balanced,0.31782400608062744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,4,1,power_law_1.2,1.627225685119629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,8,balanced,0.323525329430898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,8,balanced,0.3402453263600667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,8,balanced,0.3677599827448527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,4,1,power_law_1.2,2.1185407638549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,4,1,power_law_1.01,2.534547233581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,4,1,power_law_1.2,1.9544191360473633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,4,1,power_law_1.01,8.597382354736329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,4,1,power_law_1.2,2.5191743850708006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,8,balanced,0.2584000031153361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,4,1,power_law_1.01,3.4119552612304687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,8,balanced,0.2589226762453715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,8,balanced,0.2643146713574727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,8,balanced,0.27427732944488525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,4,1,power_law_1.2,1.8151679992675782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,8,balanced,0.37335999806722003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,8,balanced,0.37484800815582275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,8,balanced,0.4148000081380208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,8,balanced,0.468448003133138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,8,balanced,0.5983733336130778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,4,1,power_law_1.2,2.0045759201049806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,8,balanced,0.7299040158589681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,4,1,power_law_1.01,4.267916870117188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,8,balanced,1.0907893180847168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,4,1,power_law_1.2,2.400147247314453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,8,balanced,0.28673599163691205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,8,balanced,0.30034132798512775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,8,balanced,0.30364267031351727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,8,balanced,0.3189120093981425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,8,balanced,0.3162026604016622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,8,balanced,0.3704746564229329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,4,1,power_law_1.01,12.659782409667969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,8,balanced,0.43292800585428876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,4,1,power_law_1.2,1.517529582977295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,4,1,power_law_1.01,6.001273727416992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,8,balanced,1.4578293164571126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,4,1,power_law_1.2,1.734899139404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,8,balanced,2.3681599299112954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,4,1,power_law_1.2,2.1691263198852537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,4,1,power_law_1.2,2.8264575958251954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,8,balanced,0.5717440048853556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,8,balanced,3.2206827799479165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,8,balanced,0.7125333150227865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,8,balanced,0.044010668992996216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,4,1,power_law_1.2,2.587443161010742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,8,balanced,1.0773226420084636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,4,1,power_law_1.2,3.65978889465332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,8,balanced,5.138762791951497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,8,balanced,0.12244266271591187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,8,balanced,0.12552000085512796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,4,1,power_law_1.2,3.45428466796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,4,1,power_law_1.01,7.70745620727539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,8,balanced,0.04189866781234741
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,8,balanced,0.041797334949175514
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,8,balanced,0.041738669077555336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,8,balanced,0.04218666752179464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,8,balanced,0.04668800036112467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,8,balanced,1.3820160230000813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,4,1,power_law_1.01,16.670509338378906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,8,balanced,7.0559037526448565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,4,1,power_law_1.2,4.312748718261719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,8,balanced,2.024890740712484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,8,balanced,0.14114133516947427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,8,balanced,0.1430346667766571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,8,balanced,0.1455893317858378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,8,balanced,0.14553067088127136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,8,balanced,0.1471999982992808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,8,balanced,2.7471466064453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,8,balanced,0.14868266383806863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,8,balanced,0.1572640041510264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,8,balanced,0.15662933389345804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,4,1,power_law_1.2,4.726035308837891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,8,balanced,0.1616213321685791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,8,balanced,0.1790293256441752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,8,balanced,0.05625600119431814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,8,balanced,0.18518932660420737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,8,balanced,0.19508800903956094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,8,balanced,0.22699199120203653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,4,1,power_law_1.2,6.035993576049805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,8,balanced,0.25601067145665485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,8,balanced,0.2563893397649129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,8,balanced,4.483839988708496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,8,balanced,0.3223466674486796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,8,power_law_1.01,0.1428928017616272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,8,power_law_1.01,0.16332800388336183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,4,1,power_law_1.01,11.197478485107421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,4,1,power_law_1.2,6.5426177978515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,8,balanced,0.05694933235645294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,8,balanced,0.15711466471354166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,8,balanced,0.1681600014368693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,8,balanced,6.580933252970378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,8,balanced,0.15373333295186362
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,8,balanced,0.153738667567571
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,8,balanced,0.1377226710319519
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,8,balanced,0.13081066807111105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,8,balanced,0.11173866192499797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,8,balanced,0.3731946547826131
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,8,balanced,0.07052266597747803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,4,1,power_law_1.2,7.7666175842285154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,8,balanced,0.07066133121649425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,8,power_law_1.01,0.2234560012817383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,8,balanced,0.08530666430791219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,8,balanced,0.09106133381525676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,8,power_law_1.01,0.23971199989318848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,8,power_law_1.01,0.24058880805969238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,8,power_law_1.01,0.24423680305480958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,8,power_law_1.01,0.24380800724029542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,8,power_law_1.01,0.2587712049484253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,8,power_law_1.01,0.25966720581054686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,4,1,power_law_1.2,8.555359649658204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,8,power_law_1.01,0.2607487916946411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,8,power_law_1.01,0.27066879272460936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,8,power_law_1.01,0.2938240051269531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,8,power_law_1.01,0.27730560302734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,8,power_law_1.01,0.2958847999572754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,8,power_law_1.01,0.292960000038147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,8,power_law_1.01,0.2936320066452026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,8,power_law_1.01,0.29631359577178956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,8,power_law_1.01,0.3016319990158081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,4,1,power_law_1.01,14.66417236328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,8,power_law_1.01,0.2963648080825806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,8,power_law_1.01,0.3291584014892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,8,power_law_1.01,0.29802238941192627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,4,1,power_law_1.2,11.264729309082032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,8,power_law_1.01,0.30921599864959715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,8,power_law_1.01,0.30751359462738037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,8,power_law_1.01,0.3787839889526367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,8,balanced,0.49214935302734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,8,power_law_1.01,0.33665919303894043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,8,balanced,0.1111893355846405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,8,power_law_1.01,0.3442879915237427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,8,balanced,0.1325759987036387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,8,balanced,0.1797920068105062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,8,power_law_1.01,0.3573440074920654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,8,power_law_1.01,0.36061439514160154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,8,power_law_1.01,0.3869312047958374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,4,1,power_law_1.2,12.755999755859374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,8,power_law_1.01,0.40787839889526367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,8,balanced,0.6193493207295736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,8,power_law_1.01,0.40825600624084474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,8,power_law_1.01,0.43690881729125974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,8,balanced,0.936288038889567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,8,balanced,0.22105600436528525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,8,power_law_1.01,0.4249152183532715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,8,power_law_1.01,0.46654720306396485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,8,balanced,0.31201066573460895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,8,balanced,1.1852586269378662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,8,balanced,0.40692798296610516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,8,balanced,0.5925653378168741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,8,balanced,1.7549813588460286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,8,balanced,0.773845354715983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,8,power_law_1.01,0.6706624031066895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,8,balanced,2.3320159912109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,8,power_law_1.01,0.6041664123535156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,8,power_law_1.01,0.7959807872772217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,8,balanced,3.469546635945638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,8,balanced,4.613685290018718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,1,8,power_law_1.2,0.16236159801483155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,1,8,power_law_1.2,0.28615679740905764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,1,8,power_law_1.2,0.2943295955657959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,8,power_law_1.01,1.1641728401184082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,1,8,power_law_1.2,0.29674880504608153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,1,8,power_law_1.2,0.296012806892395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,1,8,power_law_1.2,0.2947200059890747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,1,8,power_law_1.2,0.3001215934753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,1,8,power_law_1.2,0.3011840105056763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,8,power_law_1.01,0.8520256042480469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,1,8,power_law_1.2,0.32279040813446047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,1,8,power_law_1.2,0.3407936096191406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,1,8,power_law_1.2,0.14476159811019898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,1,8,power_law_1.2,0.3435391902923584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,1,8,power_law_1.2,0.22997119426727294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,8,power_law_1.01,1.2537343978881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,1,8,power_law_1.2,0.2377216100692749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,8,power_law_1.01,1.533459186553955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,4,1,power_law_1.2,14.710809326171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,1,8,power_law_1.2,0.23887360095977783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,1,8,power_law_1.2,0.24382081031799316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,8,power_law_1.01,1.5558272361755372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,1,8,power_law_1.2,0.24651520252227782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,1,8,power_law_1.2,0.2514879941940308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,1,8,power_law_1.2,0.2701184034347534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,1,8,power_law_1.2,0.26991360187530516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,1,8,power_law_1.2,0.273471999168396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,1,8,power_law_1.2,0.35570559501647947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,1,8,power_law_1.2,0.2720383882522583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,1,8,power_law_1.2,0.3592128038406372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,1,8,power_law_1.2,0.2879168033599854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,1,8,power_law_1.2,0.30624001026153563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,1,8,power_law_1.2,0.3880192041397095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,8,power_law_1.01,2.314566421508789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,1,8,power_law_1.2,0.3426111936569214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,1,8,power_law_1.2,0.41276798248291013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,1,8,power_law_1.2,0.3772032022476196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,1,8,power_law_1.2,0.4466879844665527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,1,8,power_law_1.2,0.4060671806335449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,1,8,power_law_1.2,0.5091328144073486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,8,power_law_1.01,2.7857664108276365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,1,8,power_law_1.2,0.4563583850860596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,8,power_law_1.01,2.342310333251953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,1,8,power_law_1.2,0.6902016162872314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,4,1,power_law_1.2,16.37317810058594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,8,power_law_1.01,0.12038400173187255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,8,power_law_1.01,0.12511999607086183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,1,8,power_law_1.2,0.6687871932983398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,8,power_law_1.01,0.1410752058029175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,8,power_law_1.01,0.14338560104370118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,8,power_law_1.01,0.13722879886627198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,8,power_law_1.01,0.14187519550323485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,8,power_law_1.01,0.0412416011095047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,8,power_law_1.01,0.14599679708480834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,8,power_law_1.01,0.04170880019664765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,8,power_law_1.01,0.1522112011909485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,1,8,power_law_1.2,0.809062385559082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,8,power_law_1.01,0.1571328043937683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,8,power_law_1.01,0.1637312054634094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,8,power_law_1.01,0.17976959943771362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,1,8,power_law_1.2,0.8143679618835449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,8,power_law_1.01,0.19249279499053956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,8,power_law_1.01,0.20390400886535645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,8,power_law_1.01,0.23462400436401368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,8,power_law_1.01,0.2642496109008789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,8,power_law_1.01,0.04161919951438904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,8,power_law_1.01,0.04131200015544891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,8,power_law_1.01,0.29373440742492674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,8,power_law_1.01,0.042854401469230655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,1,8,power_law_1.2,1.2785152435302733
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,8,power_law_1.01,0.04699519872665405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,8,power_law_1.01,0.3317055940628052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,8,power_law_1.01,0.054201602935791016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,8,power_law_1.01,0.05431680083274841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,8,power_law_1.01,0.5125247955322265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,8,power_law_1.01,0.10931839942932128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,8,power_law_1.01,0.09891200065612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,8,power_law_1.01,0.07048959732055664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,8,power_law_1.01,0.12354559898376465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,8,power_law_1.01,0.10442240238189697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,8,power_law_1.01,0.0735360026359558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,8,power_law_1.01,0.5889535903930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,8,power_law_1.01,0.10338560342788697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,8,power_law_1.01,0.08538240194320679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,8,power_law_1.01,3.715321731567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,1,8,power_law_1.2,1.305241584777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,8,power_law_1.01,0.09440000057220459
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,8,power_law_1.01,0.13030400276184081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,1,8,power_law_1.2,1.662099266052246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,8,power_law_1.01,0.8815872192382812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,8,power_law_1.01,0.14287999868392945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,8,power_law_1.01,0.19480960369110106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,1,8,power_law_1.2,0.12086399793624877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,1,8,power_law_1.2,0.12497279644012452
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,8,power_law_1.01,0.2611583948135376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,8,power_law_1.01,5.83504638671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,1,8,power_law_1.2,0.14252159595489503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,1,8,power_law_1.2,0.14407680034637452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,1,8,power_law_1.2,0.14407680034637452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,1,8,power_law_1.2,0.1430400013923645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,1,8,power_law_1.2,0.14382079839706421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,1,8,power_law_1.2,0.14999680519104003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,8,power_law_1.01,0.3267263889312744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,1,8,power_law_1.2,0.16547199487686157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,1,8,power_law_1.2,0.16990079879760742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,1,8,power_law_1.2,0.17950719594955444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,1,8,power_law_1.2,0.20453760623931885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,1,8,power_law_1.2,0.21050240993499755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,8,power_law_1.01,1.2081727981567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,8,power_law_1.01,4.986617660522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,1,8,power_law_1.2,1.710848045349121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,1,8,power_law_1.2,0.03909760117530823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,1,8,power_law_1.2,0.038790398836135866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,1,8,power_law_1.2,0.03895680010318756
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,1,8,power_law_1.2,0.03939839899539947
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,1,8,power_law_1.2,0.04012799859046936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,1,8,power_law_1.2,0.24417920112609864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,1,8,power_law_1.2,0.04431360065937042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,1,8,power_law_1.2,0.26326398849487304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,1,8,power_law_1.2,0.05170559883117676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,1,8,power_law_1.2,0.052211201190948485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,1,8,power_law_1.2,0.1057088017463684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,1,8,power_law_1.2,0.30151679515838625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,1,8,power_law_1.2,0.09313920140266418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,1,8,power_law_1.2,0.06466559767723083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,1,8,power_law_1.2,2.4592191696166994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,1,8,power_law_1.2,0.33327360153198243
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,1,8,power_law_1.2,0.12479360103607177
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,1,8,power_law_1.2,0.1308608055114746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,1,8,power_law_1.2,0.07056639790534973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,1,8,power_law_1.2,0.12659840583801268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,1,8,power_law_1.2,0.477459192276001
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,1,8,power_law_1.2,0.0864960014820099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,8,power_law_1.01,5.970668792724609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,1,8,power_law_1.2,0.603001594543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,8,power_law_1.01,1.6796735763549804
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,1,8,power_law_1.2,0.0940671980381012
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,1,8,power_law_1.2,0.12407040596008301
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,1,8,power_law_1.2,0.14904320240020752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,8,power_law_1.01,0.44336638450622556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,1,8,power_law_1.2,2.5693952560424806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,1,8,power_law_1.2,0.8815808296203613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,1,8,power_law_1.2,0.20193920135498047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,8,power_law_1.01,2.1242176055908204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,8,power_law_1.01,8.555744171142578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,1,8,power_law_1.2,3.26861457824707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,1,8,power_law_1.2,0.25177600383758547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,1,8,power_law_1.2,3.2890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,1,8,power_law_1.2,1.2123519897460937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,8,power_law_1.01,3.326233673095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,1,8,power_law_1.2,0.35180160999298093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,8,power_law_1.01,0.6746687889099121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,8,power_law_1.01,12.751628875732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,1,8,power_law_1.2,0.4685567855834961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,1,8,power_law_1.2,1.7557247161865235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,8,power_law_1.01,12.343014526367188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,8,power_law_1.01,4.204998397827149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,1,8,power_law_1.2,5.860134506225586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,8,power_law_1.01,0.8847999572753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,1,8,power_law_1.2,2.3346879959106444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,1,8,power_law_1.2,4.893056106567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,8,power_law_1.01,1.2057408332824706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,1,8,power_law_1.2,0.6867584228515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,8,power_law_1.01,6.726316833496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,1,8,power_law_1.2,8.796345520019532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,1,8,power_law_1.2,6.597612762451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,1,8,power_law_1.2,3.496556854248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,1,8,power_law_1.2,0.9014399528503418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,8,power_law_1.01,16.659315490722655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,8,1,balanced,0.05177066723505656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,8,1,balanced,0.06580266853173573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,8,1,balanced,0.08683733145395915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,8,1,balanced,0.08720533053080241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,8,1,balanced,0.11607999602953593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,8,1,balanced,0.11705600221951802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,8,1,balanced,0.11775466799736023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,8,1,balanced,0.12029332915941875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,8,1,balanced,0.11960533261299133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,8,1,balanced,0.11945066849390666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,8,1,balanced,0.12105600039164226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,8,1,balanced,0.1279039978981018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,8,1,balanced,0.12930666406949362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,8,1,balanced,0.12963733077049255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,8,1,balanced,0.1504533290863037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,8,1,balanced,0.15518400073051453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,8,1,balanced,0.15763200322786966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,8,1,balanced,0.22739734252293906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,8,1,balanced,0.2540000081062317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,8,1,balanced,0.360586682955424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,8,1,balanced,0.4602026542027791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,8,1,balanced,0.643178661664327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,8,1,balanced,0.8403519789377848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,8,1,balanced,1.230784018834432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,8,1,balanced,1.6260053316752117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,8,1,balanced,2.5334879557291665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,8,1,balanced,3.2971200942993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,8,1,balanced,0.03756266583998998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,8,1,balanced,0.05820799867312113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,8,1,balanced,0.07056533296902974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,8,1,balanced,0.07442666590213776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,8,1,balanced,0.08960533142089844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,8,1,balanced,0.09196266531944275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,8,1,balanced,0.0930613378683726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,8,1,balanced,0.09475200374921162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,8,1,balanced,0.09270933270454407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,8,1,balanced,0.09686932961146037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,8,1,balanced,0.097461332877477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,8,1,balanced,0.10062932968139648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,8,1,balanced,0.10851200421651204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,8,1,balanced,0.11430399616559346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,8,1,balanced,0.1549013356367747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,8,1,balanced,0.16195199886957803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,8,1,balanced,0.1674506664276123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,8,1,balanced,0.21222933133443198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,8,1,balanced,0.2582079966862996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,8,1,balanced,0.35540799299875897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,8,1,balanced,0.45585068066914874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,8,1,balanced,0.6485493183135986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,8,1,balanced,0.8428053061167399
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,8,power_law_1.01,1.7320640563964844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,8,1,balanced,0.043706665436426796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,8,1,balanced,1.2367573579152424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,8,1,balanced,0.05611200133959452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,8,1,balanced,0.06818133095900218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,8,1,balanced,0.06959466636180878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,8,1,balanced,0.06863999863465627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,8,1,balanced,0.06965333223342896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,8,1,balanced,1.6347626050313313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,8,1,balanced,0.07072533170382182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,8,1,balanced,0.07087466617425282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,8,1,balanced,0.07236800094445546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,8,1,balanced,0.07457600037256877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,8,1,balanced,0.07513066629568736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,8,1,balanced,0.08063999811808269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,8,1,balanced,0.08158400158087413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,8,1,balanced,2.4243733088175454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,8,1,balanced,3.21562131245931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,8,1,balanced,0.09081600109736125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,8,1,balanced,0.12558399637540182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,8,1,balanced,0.1274773379166921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,8,1,balanced,0.12851732969284058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,8,1,balanced,0.17100266615549722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,8,1,balanced,0.21161067485809326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,8,1,balanced,0.287173330783844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,8,1,balanced,0.3615146478017171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,8,1,balanced,0.5166986783345541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,8,1,balanced,0.6640106836954752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,8,1,balanced,0.9687039852142334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,8,power_law_1.01,9.261209869384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,8,1,balanced,1.2713599999745686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,8,1,power_law_1.01,0.052019202709198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,8,1,power_law_1.01,0.06229760050773621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,8,1,balanced,1.8752320607503254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,8,1,power_law_1.01,0.06600959897041321
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,8,1,balanced,0.02779199928045273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,8,1,power_law_1.01,0.08456959724426269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,8,1,power_law_1.01,0.11341439485549927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,8,1,power_law_1.01,0.11397119760513305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,8,1,balanced,2.481680075327555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,8,1,power_law_1.01,0.11534080505371094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,8,1,power_law_1.01,0.11664639711380005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,8,1,power_law_1.01,0.11867519617080688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,8,1,power_law_1.01,0.12067840099334717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,8,1,power_law_1.01,0.12343679666519165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,8,1,power_law_1.01,0.12888959646224976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,8,1,power_law_1.01,0.13257600069046022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,8,1,power_law_1.01,0.14128639698028564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,8,1,balanced,0.029653333127498627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,8,1,balanced,0.0395413339138031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,8,1,balanced,0.03945599993069967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,8,1,balanced,0.039813332259655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,8,1,balanced,0.04058666775623957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,8,1,power_law_1.01,0.14890880584716798
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,8,1,balanced,0.04769066472848257
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,8,1,balanced,0.0484746644894282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,8,1,balanced,0.050399998823801674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,8,1,balanced,0.05012266834576925
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,8,1,balanced,0.05014933149019877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,8,1,power_law_1.01,0.1649664044380188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,8,1,balanced,0.05434666574001312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,8,1,balanced,0.05593599875768026
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,8,1,balanced,0.056554665168126426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,8,1,balanced,0.08321600159009297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,8,1,balanced,0.07157333195209503
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,8,1,balanced,0.06443200012048085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,8,1,balanced,0.11760000387827556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,8,1,balanced,0.08641599615414937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,8,1,balanced,0.10039466619491577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,8,1,power_law_1.01,0.212172794342041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,8,1,balanced,0.12490133444468181
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,8,1,balanced,0.16296000281969705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,8,1,power_law_1.01,0.24432640075683593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,8,1,balanced,0.204858660697937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,8,1,balanced,0.29253333806991577
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,8,1,balanced,0.3802880048751831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,8,1,power_law_1.01,0.04472320079803467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,8,1,balanced,0.550490657488505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,8,1,balanced,0.7142559687296549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,8,1,power_law_1.01,0.049747198820114136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,8,1,power_law_1.01,0.05364480018615723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,8,1,power_law_1.01,0.06000000238418579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,8,1,power_law_1.01,0.06638079881668091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,8,1,power_law_1.01,0.06628479957580566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,8,1,power_law_1.01,0.06756479740142822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,8,1,power_law_1.01,0.06811519861221313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,8,1,power_law_1.01,0.2950527906417847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,8,1,power_law_1.01,0.0559935986995697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,8,1,power_law_1.01,0.06974080204963684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,8,1,power_law_1.01,0.0592960000038147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,8,1,power_law_1.01,0.06964480280876159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,8,1,power_law_1.01,0.06941440105438232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,8,1,power_law_1.01,0.0720255970954895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,8,1,power_law_1.01,0.0865343987941742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,8,1,power_law_1.01,0.38962559700012206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,8,1,power_law_1.01,0.07390080094337463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,8,power_law_1.01,19.53008575439453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,8,1,power_law_1.01,0.09237120151519776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,1,8,power_law_1.2,4.513100814819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,8,1,power_law_1.01,0.10464639663696289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,8,1,power_law_1.01,0.11381759643554687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,8,1,power_law_1.01,0.12128640413284301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,8,1,power_law_1.01,0.4875840187072754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,8,1,power_law_1.01,0.12740479707717894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,8,1,power_law_1.01,0.1344063997268677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,8,1,power_law_1.01,0.088128000497818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,8,1,power_law_1.01,0.0883903980255127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,8,1,power_law_1.01,0.0896127998828888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,8,1,power_law_1.01,0.17690240144729613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,8,1,power_law_1.01,0.09425280094146729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,8,1,power_law_1.01,0.6760511875152588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,8,1,power_law_1.01,0.0964352011680603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,8,1,power_law_1.01,0.10788480043411255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,8,1,power_law_1.01,0.11398400068283081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,8,1,power_law_1.01,0.12532479763031007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,8,1,power_law_1.01,0.1411520004272461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,8,1,power_law_1.01,0.1488767981529236
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,8,1,power_law_1.01,0.02717440128326416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,8,1,power_law_1.01,0.029369598627090453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,8,1,power_law_1.01,0.029900801181793214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,8,1,power_law_1.01,0.21928319931030274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,8,1,power_law_1.01,0.035897600650787356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,8,1,power_law_1.01,0.03802880048751831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,8,1,power_law_1.01,0.043609601259231565
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,8,1,power_law_1.01,0.04668799936771393
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,8,1,power_law_1.01,0.04958719909191132
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,8,1,power_law_1.01,0.04855040013790131
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,8,1,power_law_1.01,0.050406402349472045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,8,1,power_law_1.01,0.05144960284233093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,8,1,power_law_1.01,0.0536191999912262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,8,1,power_law_1.01,0.1717695951461792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,8,1,power_law_1.01,0.05610880255699158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,8,1,power_law_1.01,0.18632320165634156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,8,1,power_law_1.01,0.05961599946022034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,8,1,power_law_1.01,0.06107519865036011
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,8,1,power_law_1.01,0.06550400257110596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,8,1,power_law_1.01,0.06749439835548401
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,8,1,power_law_1.01,0.2880959987640381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,1,8,power_law_1.2,11.352012634277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,8,1,power_law_1.01,0.08225280046463013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,8,1,power_law_1.01,0.3732095956802368
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,8,1,power_law_1.01,0.08894079923629761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,8,1,power_law_1.01,0.236844801902771
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,8,1,power_law_1.01,0.11111040115356445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,8,1,power_law_1.01,0.28342399597167967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,8,1,power_law_1.01,0.8675456047058105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,8,1,power_law_1.01,0.38405120372772217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,8,1,power_law_1.01,1.2673727989196777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,8,1,power_law_1.01,0.4791103839874268
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,8,1,power_law_1.01,0.12283519506454468
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,8,1,power_law_1.01,0.16984319686889648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,1,8,power_law_1.2,14.055206298828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,8,1,power_law_1.01,0.6746560096740722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,8,1,power_law_1.01,0.5228159904479981
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,8,1,power_law_1.01,0.20991361141204834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,8,1,power_law_1.01,1.6593984603881835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,8,1,power_law_1.01,0.6767360210418701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,1,8,power_law_1.2,1.3046079635620118
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,8,1,power_law_1.01,0.29663360118865967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,8,1,power_law_1.01,0.8747455596923828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,8,1,power_law_1.01,1.2843392372131348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,8,1,power_law_1.01,0.37920639514923093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,8,1,power_law_1.01,0.981875228881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,1,8,power_law_1.2,17.182406616210937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,8,1,power_law_1.01,0.549177598953247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,8,1,power_law_1.01,1.6797183990478515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,8,1,power_law_1.2,0.05240319967269898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,1,8,power_law_1.2,6.7078399658203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,8,1,power_law_1.01,2.453152084350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,1,8,power_law_1.2,20.5712890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,8,1,power_law_1.2,0.060121601819992064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,8,1,power_law_1.2,0.06535680294036865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,8,1,power_law_1.2,0.049958398938179015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,8,1,power_law_1.2,0.054476797580718994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,8,1,power_law_1.01,2.4763456344604493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,8,1,power_law_1.2,0.06120319962501526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,8,1,power_law_1.2,0.06880000233650208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,8,1,power_law_1.2,0.0864512026309967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,8,1,power_law_1.2,0.0870464026927948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,8,1,power_law_1.01,1.282975959777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,8,1,power_law_1.2,0.08883200287818908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,8,1,power_law_1.2,0.09093120098114013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,8,1,power_law_1.2,0.09258880019187928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,8,1,power_law_1.2,0.09649919867515563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,8,1,power_law_1.2,0.10782719850540161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,8,1,power_law_1.2,0.11708799600601197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,8,1,power_law_1.2,0.12616959810256959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,8,1,power_law_1.2,0.14207359552383422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,8,1,power_law_1.2,0.15534080266952516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,8,1,power_law_1.01,3.248499298095703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,8,1,power_law_1.2,0.04734080135822296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,8,1,power_law_1.2,0.054611200094223024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,8,1,power_law_1.2,0.061791998147964475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,8,1,power_law_1.2,0.06664320230484008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,8,1,power_law_1.2,0.06591359972953796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,8,1,power_law_1.2,0.06744319796562195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,8,1,power_law_1.2,0.06862080097198486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,8,1,power_law_1.2,0.06958079934120179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,8,1,power_law_1.2,0.06959999799728393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,8,1,power_law_1.2,0.07141119837760926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,8,1,power_law_1.2,0.0828607976436615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,8,1,power_law_1.2,0.07986559867858886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,8,1,power_law_1.2,0.09768319725990296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,8,1,power_law_1.2,0.17541760206222534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,8,1,power_law_1.2,0.09818879961967468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,8,1,power_law_1.2,0.11302399635314941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,8,1,power_law_1.2,0.18461439609527588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,8,1,power_law_1.2,0.23875839710235597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,8,1,power_law_1.2,0.12181119918823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,1,8,power_law_1.2,1.7693183898925782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,8,1,power_law_1.2,0.12668800354003906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,8,1,power_law_1.2,0.13447680473327636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,8,1,power_law_1.2,0.27338879108428954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,8,1,power_law_1.2,0.17860480546951293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,8,1,power_law_1.2,0.2173504114151001
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,8,1,power_law_1.2,0.02460159957408905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,8,1,power_law_1.2,0.3818048000335693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,8,1,power_law_1.2,0.027008000016212463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,8,1,power_law_1.2,0.03051519989967346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,8,1,power_law_1.2,0.03603839874267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,8,1,power_law_1.2,0.0363072007894516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,8,1,power_law_1.2,0.041555199027061465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,8,1,power_law_1.2,0.04426240026950836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,8,1,power_law_1.2,0.04819200038909912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,8,1,power_law_1.2,0.11436799764633179
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,8,1,power_law_1.2,0.04467200040817261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,8,1,power_law_1.2,0.11519999504089355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,8,1,power_law_1.2,0.047968000173568726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,8,1,power_law_1.2,0.11608959436416626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,8,1,power_law_1.2,0.11681280136108399
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,8,1,power_law_1.2,0.051846402883529666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,8,1,power_law_1.2,0.12033280134201049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,8,1,power_law_1.2,0.12005759477615356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,8,1,power_law_1.2,0.051667201519012454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,8,1,power_law_1.2,0.12281600236892701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,8,1,power_law_1.2,0.12893439531326295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,8,1,power_law_1.2,0.13317760229110717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,8,1,power_law_1.2,0.14140160083770753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,8,1,power_law_1.2,0.14950400590896606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,8,1,power_law_1.2,0.1760319948196411
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,8,1,power_law_1.2,0.053913599252700804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,8,1,power_law_1.2,0.28742399215698244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,8,1,power_law_1.2,0.21030399799346924
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,8,1,power_law_1.2,0.0585919976234436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,8,1,power_law_1.2,0.48616957664489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,8,1,power_law_1.2,0.05974400043487549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,8,1,power_law_1.2,0.2467519998550415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,8,1,power_law_1.2,0.06275839805603027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,8,1,power_law_1.01,3.2628990173339845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,8,balanced,0.2140480081240336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,8,balanced,0.3657706578572591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,8,1,power_law_1.2,0.2858815908432007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,8,balanced,0.39163732528686523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,8,1,power_law_1.2,0.06684160232543945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,8,balanced,0.3822133143742879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,8,balanced,0.3730986515680949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,8,balanced,0.389413317044576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,8,balanced,0.3725759983062744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,8,balanced,0.40043731530507404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,8,1,power_law_1.2,0.3878400087356567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,8,balanced,0.37642133235931396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,8,balanced,0.3887840112050374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,8,balanced,0.4018133481343587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,8,balanced,0.39608001708984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,8,1,power_law_1.2,0.3724479913711548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,8,1,power_law_1.2,0.47803521156311035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,8,1,power_law_1.2,0.07874559760093688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,8,balanced,0.4078559875488281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,8,balanced,0.43588801225026447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,8,balanced,0.473311980565389
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,8,1,power_law_1.2,0.0913536012172699
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,8,1,power_law_1.2,0.11270400285720825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,8,1,power_law_1.2,0.5227968215942382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,8,balanced,0.4851679801940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,8,balanced,0.48634131749471027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,8,1,power_law_1.2,0.6774015903472901
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,8,1,power_law_1.2,0.6742208003997803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,8,balanced,0.5285866657892863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,8,balanced,0.6030400196711222
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,8,1,power_law_1.2,0.13363200426101685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,8,balanced,0.7845119635264078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,8,balanced,0.97762664159139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,8,balanced,1.594538688659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,8,balanced,2.229637304941813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,8,balanced,3.6224212646484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,8,balanced,5.030469258626302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,8,1,power_law_1.2,0.6758463859558106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,8,balanced,7.733114878336589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,8,1,power_law_1.01,1.8916095733642577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,8,balanced,10.597818374633789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,8,balanced,0.18677866458892822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,8,balanced,0.28294400374094647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,8,balanced,0.3016960024833679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,8,balanced,0.29452266295750934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,8,balanced,0.299615999062856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,8,balanced,0.3022986650466919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,8,balanced,0.3038826584815979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,8,balanced,0.31033066908518475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,8,balanced,0.31683733065923053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,8,balanced,0.31681599219640094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,8,balanced,0.31887465715408325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,8,balanced,0.33319467306137085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,8,balanced,0.35066668192545575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,8,1,power_law_1.2,0.884006404876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,8,balanced,0.37193067868550617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,8,balanced,0.38066665331522626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,8,balanced,0.3949600060780843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,8,1,power_law_1.2,0.16910719871520996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,8,balanced,0.40238932768503827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,8,balanced,0.48348267873128253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,8,balanced,0.5723413228988647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,8,1,power_law_1.2,0.8818367958068848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,8,balanced,0.7747253576914469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,8,balanced,1.0111520290374756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,8,balanced,1.4909706115722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,8,balanced,1.9214240709940593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,8,balanced,2.8670454025268555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,8,balanced,4.051381429036458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,8,1,power_law_1.2,0.20983679294586183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,8,balanced,6.783728281656901
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,1,8,power_law_1.2,9.031053161621093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,8,balanced,0.1495199998219808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,8,balanced,0.16552000244458517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,8,balanced,0.17320533593495688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,8,balanced,9.903717041015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,8,1,power_law_1.2,1.276972770690918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,8,balanced,0.17606399456659952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,8,balanced,0.17653866608937582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,8,balanced,0.17709332704544067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,8,balanced,0.18362667163213095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,8,balanced,0.18619734048843384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,8,balanced,0.19178666671117148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,8,balanced,0.19493865966796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,8,balanced,0.2031946579615275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,8,balanced,0.21711466709772745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,8,balanced,0.22393600145975748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,8,balanced,0.24672534068425497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,8,balanced,0.27994134028752643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,8,balanced,0.2998239994049072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,8,balanced,0.3244853417078654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,8,balanced,0.40075735251108807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,8,balanced,0.46833598613739014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,8,1,power_law_1.2,0.9780351638793945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,8,balanced,0.6287039915720621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,8,balanced,0.8585440317789713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,8,balanced,1.1981066862742107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,8,balanced,1.5907413164774578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,8,balanced,2.353818734486898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,8,1,power_law_1.2,1.2795136451721192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,8,balanced,3.126469294230143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,8,1,power_law_1.2,1.6642112731933594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,8,balanced,4.667706807454427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,8,balanced,6.202725092569987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,8,power_law_1.01,0.21203200817108153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,8,power_law_1.01,0.3578304052352905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,8,power_law_1.01,0.3782399892807007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,8,power_law_1.01,0.38226559162139895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,8,power_law_1.01,0.3683135986328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,8,power_law_1.01,0.3818432092666626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,8,power_law_1.01,0.3840895891189575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,8,power_law_1.01,0.39913599491119384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,8,1,power_law_1.01,2.505670356750488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,8,power_law_1.01,0.4207104206085205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,8,power_law_1.01,0.18440959453582764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,8,power_law_1.01,0.417087984085083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,8,power_law_1.01,0.2760832071304321
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,8,1,power_law_1.01,0.7268095970153808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,8,power_law_1.01,0.2930239915847778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,8,power_law_1.01,0.2936448097229004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,8,power_law_1.01,0.29363839626312255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,8,power_law_1.01,0.30478720664978026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,8,power_law_1.01,0.31002240180969237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,8,power_law_1.01,0.31956479549407957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,8,power_law_1.01,0.341919994354248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,8,power_law_1.01,0.34182400703430177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,8,power_law_1.01,0.4459712028503418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,8,power_law_1.01,0.34640638828277587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,8,power_law_1.01,0.45511679649353026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,8,power_law_1.01,0.3667392015457153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,8,power_law_1.01,0.4772223949432373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,8,power_law_1.01,0.39913599491119384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,8,power_law_1.01,0.49777917861938475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,8,power_law_1.01,0.43038082122802734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,8,power_law_1.01,0.5701824188232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,8,power_law_1.01,0.5185984134674072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,8,power_law_1.01,0.5406144142150879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,8,power_law_1.01,0.630783987045288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,8,power_law_1.01,0.6881728172302246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,8,power_law_1.01,0.7011072158813476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,8,balanced,0.05862933397293091
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,8,balanced,0.05826666454474131
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,8,balanced,0.05823466678460439
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,8,balanced,0.0583840012550354
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,8,balanced,0.058090666929880776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,8,power_law_1.01,1.089023971557617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,8,balanced,0.06264000137646993
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,8,balanced,0.0710453341404597
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,8,balanced,0.08999466896057129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,8,balanced,0.2151306668917338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,8,balanced,0.20853867133458456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,8,1,power_law_1.2,1.2904704093933106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,8,power_law_1.01,1.0734335899353027
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,8,balanced,0.19606933991114298
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,8,balanced,0.1839146614074707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,8,balanced,0.18100800116856894
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,8,balanced,0.15987199544906616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,8,balanced,0.13934399684270224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,8,balanced,0.08275733391443889
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,8,balanced,0.08673600355784099
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,8,balanced,0.11031466722488403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,8,power_law_1.01,1.2265536308288574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,8,balanced,0.11754666765530904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,8,balanced,0.14758933583895364
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,8,balanced,0.18945600589116415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,8,balanced,0.25124265750249225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,8,balanced,0.32474132378896076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,8,power_law_1.01,1.6540544509887696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,8,balanced,0.46851734320322674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,8,power_law_1.01,1.303225612640381
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,8,balanced,0.6120800177256266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,8,balanced,0.9036533037821451
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,8,balanced,1.206496000289917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,8,1,power_law_1.2,0.29322240352630613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,8,power_law_1.01,1.9755071640014648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,8,power_law_1.01,2.513875198364258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,8,1,power_law_1.2,1.6608064651489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,8,power_law_1.01,4.006796646118164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,8,power_law_1.01,2.5745920181274413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,8,power_law_1.01,3.925939178466797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,8,power_law_1.01,5.071180725097657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,8,power_law_1.01,0.05694079995155334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,8,power_law_1.01,0.056831997632980344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,8,power_law_1.01,0.05546879768371582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,8,power_law_1.01,0.05604479908943176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,8,power_law_1.01,0.055820798873901366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,8,power_law_1.01,0.060089600086212155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,8,power_law_1.01,0.06691200137138367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,8,power_law_1.01,0.07022719979286193
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,8,power_law_1.01,0.12767360210418702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,8,power_law_1.01,0.10458240509033204
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,8,power_law_1.01,0.073471999168396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,8,power_law_1.01,0.12561919689178466
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,8,power_law_1.01,0.17373440265655518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,8,power_law_1.01,5.805177688598633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,8,power_law_1.01,0.09515519738197327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,8,power_law_1.01,0.14420479536056519
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,8,power_law_1.01,0.11413760185241699
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,8,power_law_1.01,0.12201600074768067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,8,1,power_law_1.2,0.3866048097610474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,8,power_law_1.01,0.1741376042366028
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,8,power_law_1.01,0.21517438888549806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,8,power_law_1.01,0.30831360816955566
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,8,power_law_1.01,0.41887998580932617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,8,1,power_law_1.2,1.8936384201049805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,8,power_law_1.01,0.5869887828826904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,8,power_law_1.01,7.662700653076172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,8,1,power_law_1.2,0.5512959957122803
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,8,power_law_1.01,0.8034239768981933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,8,power_law_1.01,9.790214538574219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,8,power_law_1.01,1.1049344062805175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,8,1,power_law_1.2,2.4721216201782226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,8,power_law_1.01,11.701696014404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,8,power_law_1.01,1.5280256271362305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,8,power_law_1.01,14.146284484863282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,8,1,power_law_1.2,2.4550336837768554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,8,power_law_1.01,0.1486847996711731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,8,power_law_1.01,0.1645311951637268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,8,power_law_1.01,0.1712448000907898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,8,power_law_1.01,0.17455359697341918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,8,power_law_1.01,0.17357439994812013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,8,power_law_1.01,0.17597440481185914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,8,power_law_1.01,0.18110079765319825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,8,power_law_1.01,0.18691840171813964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,8,power_law_1.01,0.20229120254516603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,8,power_law_1.01,0.20869119167327882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,8,power_law_1.01,0.2226815938949585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,8,power_law_1.01,0.2457024097442627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,8,power_law_1.01,0.2690432071685791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,8,power_law_1.01,0.3083775997161865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,8,power_law_1.01,0.34565119743347167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,8,power_law_1.01,0.38757760524749757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,8,power_law_1.01,0.5014463901519776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,8,power_law_1.01,0.6915775775909424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,8,power_law_1.01,0.803212833404541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,8,power_law_1.01,1.1698431968688965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,8,power_law_1.01,1.778950309753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,8,power_law_1.01,2.3740352630615233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,8,power_law_1.01,3.2340415954589843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,8,power_law_1.01,2.417804718017578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,8,1,power_law_1.2,0.7360576152801513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,1,8,power_law_1.2,0.21070079803466796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,1,8,power_law_1.2,0.3573760032653809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,1,8,power_law_1.2,0.3803776025772095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,1,8,power_law_1.2,0.37960960865020754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,1,8,power_law_1.2,0.3816447973251343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,1,8,power_law_1.2,0.3913536071777344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,1,8,power_law_1.2,0.38777599334716795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,1,8,power_law_1.2,0.39628798961639405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,8,1,power_law_1.2,3.2287361145019533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,1,8,power_law_1.2,0.19824639558792115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,1,8,power_law_1.2,0.29685120582580565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,1,8,power_law_1.2,0.29793920516967776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,1,8,power_law_1.2,0.29531519412994384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,1,8,power_law_1.2,0.3066751956939697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,1,8,power_law_1.2,0.3014847993850708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,1,8,power_law_1.2,0.32343039512634275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,1,8,power_law_1.2,0.41244797706604003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,1,8,power_law_1.2,0.3322432041168213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,1,8,power_law_1.2,0.4341695785522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,1,8,power_law_1.2,0.34024319648742674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,1,8,power_law_1.2,0.45043201446533204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,1,8,power_law_1.2,0.3463295936584473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,1,8,power_law_1.2,0.4602687835693359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,1,8,power_law_1.2,0.3563391923904419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,1,8,power_law_1.2,0.3701632022857666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,1,8,power_law_1.2,0.4756927967071533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,1,8,power_law_1.2,0.4250688076019287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,1,8,power_law_1.2,0.5100224018096924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,1,8,power_law_1.2,0.4565120220184326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,8,1,power_law_1.2,3.241862487792969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,1,8,power_law_1.2,0.5449344158172608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,1,8,power_law_1.2,0.5244863986968994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,1,8,power_law_1.2,0.5987008094787598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,1,8,power_law_1.2,0.5649151802062988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,1,8,power_law_1.2,0.6753280162811279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,1,8,power_law_1.2,0.646227216720581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,1,8,power_law_1.2,1.0309311866760253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,8,power_law_1.01,4.519737625122071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,1,8,power_law_1.2,1.030016040802002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,1,8,power_law_1.2,1.282259178161621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,8,power_law_1.01,21.152210998535157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,1,8,power_law_1.2,1.3270400047302247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,8,power_law_1.01,24.246099853515624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,1,8,power_law_1.2,1.925222396850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,1,8,power_law_1.2,1.9311744689941406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,1,8,power_law_1.2,2.6477888107299803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,8,power_law_1.01,6.531680297851563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,1,8,power_law_1.2,2.490777587890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,1,8,power_law_1.2,0.14797439575195312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,1,8,power_law_1.2,0.163756799697876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,1,8,power_law_1.2,4.075129699707031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,1,8,power_law_1.2,0.17159680128097535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,1,8,power_law_1.2,0.17279360294342042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,1,8,power_law_1.2,0.1727552056312561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,1,8,power_law_1.2,0.17481600046157836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,1,8,power_law_1.2,3.6955520629882814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,1,8,power_law_1.2,0.18177920579910278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,1,8,power_law_1.2,0.18746880292892457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,1,8,power_law_1.2,0.20277760028839112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,8,1,power_law_1.2,2.5153791427612306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,1,8,power_law_1.2,0.20944640636444092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,1,8,power_law_1.2,0.22318079471588134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,1,8,power_law_1.2,0.25109119415283204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,1,8,power_law_1.2,0.26372480392456055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,1,8,power_law_1.2,0.3113343954086304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,1,8,power_law_1.2,4.903814315795898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,1,8,power_law_1.2,0.36589438915252687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,1,8,power_law_1.2,0.3817471981048584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,8,power_law_1.01,3.112998390197754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,8,1,balanced,0.07549866537253062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,8,1,balanced,0.10944533348083496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,8,1,balanced,0.15825066963831583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,8,1,balanced,0.16059199968973795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,8,1,balanced,0.17061867316563925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,8,1,balanced,0.17072532574335733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,1,8,power_law_1.2,0.5015935897827148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,8,1,balanced,0.17070933183034262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,8,1,balanced,0.17318934202194214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,8,1,balanced,0.1736746629079183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,8,1,balanced,0.1695573329925537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,8,1,balanced,0.1708959937095642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,8,1,balanced,0.1747573415438334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,1,8,power_law_1.2,0.7088768005371093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,1,8,power_law_1.2,0.9215295791625977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,8,1,balanced,0.17548799514770508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,8,1,balanced,0.17916800578435263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,8,1,balanced,0.23399466276168823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,8,1,balanced,0.23982399702072144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,8,1,balanced,0.2461493412653605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,8,1,balanced,0.33745598793029785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,8,1,balanced,0.3927733500798543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,8,1,balanced,0.576906681060791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,8,1,balanced,0.7176960309346517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,8,1,balanced,1.0420746803283691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,8,1,balanced,1.3529493014017742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,8,1,balanced,2.0372959772745767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,1,8,power_law_1.2,4.653184127807617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,1,8,power_law_1.2,1.2553600311279296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,8,1,balanced,2.746255874633789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,8,1,balanced,4.227018674214681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,8,1,balanced,5.754677454630534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,8,1,balanced,0.050000001986821495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,8,1,balanced,0.08291733264923096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,8,1,balanced,0.10682666301727295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,8,1,balanced,0.1125866671403249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,8,1,balanced,0.12407466769218445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,8,1,balanced,0.12607466181119284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,8,1,balanced,0.12938666343688965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,8,1,balanced,0.12838400403658548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,8,1,balanced,0.12985600034395853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,8,1,balanced,0.1306933363278707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,8,1,balanced,0.1363093356291453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,8,1,balanced,0.13784533739089966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,1,8,power_law_1.2,1.7690431594848632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,8,1,balanced,0.14924800395965576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,8,1,balanced,0.15599999825159708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,8,1,balanced,0.21670933564503989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,8,1,balanced,0.22799466053644815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,8,1,balanced,0.23146132628122965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,8,1,balanced,0.3166453242301941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,8,1,balanced,0.3863413333892822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,8,1,balanced,0.5491093397140503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,1,8,power_law_1.2,2.31329288482666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,8,1,balanced,0.708405335744222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,1,8,power_law_1.2,10.160262298583984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,8,1,balanced,1.0250879923502605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,8,1,balanced,1.3475573857625325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,8,1,balanced,1.9946880340576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,8,1,balanced,2.634415944417318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,8,1,balanced,3.9245227177937827
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,1,8,power_law_1.2,0.0537280023097992
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,1,8,power_law_1.2,0.054099202156066895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,1,8,power_law_1.2,0.054092800617218016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,1,8,power_law_1.2,0.054425597190856934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,1,8,power_law_1.2,0.05475839972496033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,8,1,balanced,5.213280042012532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,8,1,balanced,0.05978666742642721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,8,1,balanced,0.07236266632874806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,8,1,balanced,0.09902399778366089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,8,1,balanced,0.09729066491127014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,8,1,balanced,0.09830400347709656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,8,1,balanced,0.10079466303189595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,8,1,balanced,0.10104533036549886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,8,1,balanced,0.10109333197275798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,8,1,balanced,0.1035146713256836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,8,1,balanced,0.10340799887975057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,8,1,balanced,0.10583466291427612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,8,1,balanced,0.11508267124493916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,8,1,balanced,0.11680533488591512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,8,1,balanced,0.13272533814112344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,8,1,balanced,0.1673120061556498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,8,1,balanced,0.17246399323145548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,8,1,balanced,0.1800266702969869
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,1,8,power_law_1.2,0.05845119953155518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,8,1,balanced,0.23995200792948404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,8,1,balanced,0.30268800258636475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,8,1,balanced,0.42293334007263184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,8,1,balanced,0.5283573468526205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,8,1,balanced,0.7714453538258871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,8,1,balanced,0.9971093336741129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,8,1,balanced,1.473024050394694
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,1,8,power_law_1.2,0.065830397605896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,8,1,balanced,1.925370693206787
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,1,8,power_law_1.2,0.06556159853935242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,1,8,power_law_1.2,0.1278720021247864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,1,8,power_law_1.2,0.1219648003578186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,8,1,balanced,2.8696959813435874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,1,8,power_law_1.2,0.07152000069618225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,8,power_law_1.01,10.386431884765624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,1,8,power_law_1.2,0.15567359924316407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,1,8,power_law_1.2,0.11119359731674194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,8,1,balanced,3.794485410054525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,1,8,power_law_1.2,7.84755859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,1,8,power_law_1.2,0.09425920248031616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,1,8,power_law_1.2,0.1722815990447998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,1,8,power_law_1.2,0.1122048020362854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,1,8,power_law_1.2,3.49194221496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,1,8,power_law_1.2,0.13841279745101928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,1,8,power_law_1.2,0.16947200298309326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,1,8,power_law_1.2,0.22602880001068115
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,8,1,balanced,0.031370667119820915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,8,1,balanced,0.03588266670703888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,8,1,balanced,0.054527997970581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,8,1,balanced,0.05504000186920166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,8,1,balanced,0.054618666569391884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,8,1,balanced,0.05565866827964783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,8,1,balanced,0.06067200005054474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,8,1,balanced,0.061887999375661217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,8,1,balanced,0.06423466900984447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,8,1,balanced,0.06425599753856659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,8,power_law_1.01,32.13498229980469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,8,1,balanced,0.06426666676998138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,8,1,balanced,0.07045333087444305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,8,1,balanced,0.06981866558392842
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,8,1,balanced,0.07336000104745229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,8,1,balanced,0.09797333677609761
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,8,1,balanced,0.08570133646329244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,8,1,balanced,0.08467732866605122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,8,1,balanced,0.14612266421318054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,8,1,balanced,0.11143466830253601
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,8,1,balanced,0.14097600181897482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,8,1,balanced,0.16911466916402182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,8,1,balanced,0.2379680077234904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,1,8,power_law_1.2,0.30773119926452636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,8,1,balanced,0.30186132589975995
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,8,1,balanced,0.43542933464050293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,8,1,balanced,0.5667999982833862
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,8,1,balanced,0.8251573244730631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,8,1,power_law_1.01,0.07275519967079162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,8,1,power_law_1.01,0.08272640109062195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,8,1,balanced,1.0814399719238281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,8,1,power_law_1.01,0.09765120148658753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,8,1,power_law_1.01,0.06686080098152161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,8,1,power_law_1.01,0.12087039947509766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,8,1,power_law_1.01,0.07514879703521729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,8,1,power_law_1.01,0.15872000455856322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,8,1,power_law_1.01,0.08439679741859436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,8,1,power_law_1.01,0.15968639850616456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,8,1,power_law_1.01,0.10490239858627319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,8,1,power_law_1.01,0.16054400205612182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,8,1,power_law_1.01,0.11854720115661621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,8,1,power_law_1.01,0.16085760593414306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,8,1,power_law_1.01,0.12130559682846069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,8,1,power_law_1.01,0.1642624020576477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,8,1,power_law_1.01,0.12269439697265624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,8,1,power_law_1.01,0.16506240367889405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,8,1,power_law_1.01,0.12276480197906495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,8,1,power_law_1.01,0.17022080421447755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,8,1,power_law_1.01,0.13046400547027587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,8,1,power_law_1.01,0.1948799967765808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,8,1,power_law_1.01,0.1386944055557251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,8,1,power_law_1.01,0.152128005027771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,8,1,power_law_1.01,0.19648640155792235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,8,power_law_1.01,27.63457336425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,8,1,power_law_1.01,0.22042880058288575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,8,1,power_law_1.01,0.16569600105285645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,8,1,power_law_1.01,0.1788159966468811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,8,1,power_law_1.01,0.06017919778823853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,8,1,power_law_1.01,0.19969279766082765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,8,1,power_law_1.01,0.22960638999938965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,8,1,power_law_1.01,0.06753280162811279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,8,1,power_law_1.01,0.07701759934425353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,8,1,power_law_1.01,0.3014271974563599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,8,1,power_law_1.01,0.2143615961074829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,8,1,power_law_1.01,0.09496319890022278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,8,1,power_law_1.01,0.25522561073303224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,8,1,power_law_1.01,0.31059200763702394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,8,1,power_law_1.01,0.09791359901428223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,8,1,power_law_1.01,0.2696255922317505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,8,1,power_law_1.01,0.09589759707450866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,8,1,power_law_1.01,0.09905920028686524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,8,1,power_law_1.01,0.1004480004310608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,8,1,power_law_1.01,0.10216319561004639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,8,1,power_law_1.01,0.10928640365600586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,8,1,power_law_1.01,0.10503040552139283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,8,1,power_law_1.01,0.13423360586166383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,8,1,power_law_1.01,0.3558784008026123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,8,1,power_law_1.01,0.13941119909286498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,8,1,power_law_1.01,0.1536960005760193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,8,1,power_law_1.01,0.16552319526672363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,8,1,power_law_1.01,0.4220416069030762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,8,1,power_law_1.01,0.17193599939346313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,1,8,power_law_1.2,0.4328127861022949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,8,1,power_law_1.01,0.2062527894973755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,8,1,power_law_1.01,0.26031999588012694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,8,1,power_law_1.01,0.3851072072982788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,8,1,power_law_1.01,0.5839615821838379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,8,1,power_law_1.01,0.3099776029586792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,8,1,power_law_1.01,0.42899198532104493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,8,1,power_law_1.01,0.461568021774292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,8,1,power_law_1.01,0.7397952079772949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,8,1,power_law_1.01,0.6107776165008545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,1,8,power_law_1.2,0.595468807220459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,1,8,power_law_1.2,11.87737579345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,1,8,power_law_1.2,5.146470260620117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,8,1,power_law_1.01,0.7791103839874267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,8,1,power_law_1.01,0.5501887798309326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,8,1,power_law_1.01,1.0642239570617675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,1,8,power_law_1.2,15.419999694824218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,8,1,power_law_1.01,0.783616018295288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,8,1,power_law_1.01,1.3783103942871093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,8,1,power_law_1.01,1.0194239616394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,8,1,power_law_1.01,2.0267839431762695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,1,8,power_law_1.2,0.8391488075256348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,8,power_law_1.01,12.623423767089843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,8,1,power_law_1.01,1.0996543884277343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,8,1,power_law_1.01,1.4839167594909668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,8,1,power_law_1.01,0.029625600576400755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,8,1,power_law_1.01,0.03402239978313446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,8,1,power_law_1.01,0.039315199851989745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,8,1,power_law_1.01,0.052697598934173584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,8,1,power_law_1.01,0.052262401580810545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,8,1,power_law_1.01,0.05438079833984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,8,1,power_law_1.01,0.05899519920349121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,8,1,power_law_1.01,0.06422399878501892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,8,1,power_law_1.01,0.06096640229225159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,8,1,power_law_1.01,2.669856071472168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,1,8,power_law_1.2,6.680838775634766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,8,1,power_law_1.01,0.06220160126686096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,8,1,power_law_1.01,0.06630399823188782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,8,1,power_law_1.01,0.06732159852981567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,8,1,power_law_1.01,0.07030400037765502
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,8,1,power_law_1.01,0.07957760095596314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,8,1,power_law_1.01,0.07741439938545228
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,8,1,power_law_1.01,0.08260480165481568
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,8,1,power_law_1.01,0.09057279825210571
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,8,1,power_law_1.01,0.11531519889831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,8,1,power_law_1.01,1.4244864463806153
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,8,1,power_law_1.01,0.11964800357818603
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,8,1,power_law_1.01,0.1509760022163391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,1,8,power_law_1.2,1.1662655830383302
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,8,1,power_law_1.01,0.1862720012664795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,8,1,power_law_1.01,0.24751999378204345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,8,1,power_law_1.01,1.9474048614501953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,8,1,power_law_1.01,0.3166016101837158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,1,8,power_law_1.2,18.47105255126953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,8,1,power_law_1.01,0.44280319213867186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,8,1,power_law_1.01,2.0738367080688476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,1,8,power_law_1.2,1.5524736404418946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,8,1,power_law_1.01,0.5767039775848388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,8,1,power_law_1.01,2.882636833190918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,8,1,power_law_1.01,2.724582481384277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,8,1,power_law_1.01,3.939334487915039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,1,8,power_law_1.2,24.51370849609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,1,8,power_law_1.2,10.382259368896484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,8,1,power_law_1.01,3.8112255096435548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,8,1,power_law_1.2,0.07253760099411011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,8,1,power_law_1.2,0.08150399923324585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,8,1,power_law_1.2,0.09305599927902222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,8,1,power_law_1.2,0.12548480033874512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,8,1,power_law_1.2,0.15863679647445678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,8,1,power_law_1.2,0.16017919778823853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,8,1,power_law_1.2,0.16167039871215821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,8,1,power_law_1.2,0.1627135992050171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,8,1,power_law_1.2,0.16355199813842775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,8,1,power_law_1.01,0.8333951950073242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,8,1,power_law_1.2,0.16883200407028198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,8,1,power_law_1.2,0.1732800006866455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,8,1,power_law_1.2,0.19624320268630982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,8,1,power_law_1.01,4.006540679931641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,8,1,power_law_1.2,0.1992959976196289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,8,1,power_law_1.2,0.21973121166229248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,8,1,power_law_1.2,0.23057920932769777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,8,1,power_law_1.2,0.27614080905914307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,1,8,power_law_1.2,2.4563648223876955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,8,1,power_law_1.2,0.3045120000839233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,8,1,power_law_1.2,0.3801408052444458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,8,1,power_law_1.2,0.4501311779022217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,8,1,power_law_1.2,0.6201024055480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,8,1,power_law_1.2,0.7755199909210205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,8,1,power_law_1.2,1.0826560020446778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,8,1,power_law_1.2,1.411731243133545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,8,1,power_law_1.2,2.083206367492676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,1,8,power_law_1.2,30.34573974609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,8,1,power_law_1.2,0.06542720198631287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,8,1,power_law_1.2,0.07338240146636962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,8,1,power_law_1.2,0.08365439772605895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,8,1,power_law_1.2,0.10117119550704956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,8,1,power_law_1.2,0.11828479766845704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,8,1,power_law_1.2,0.1208448052406311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,8,1,power_law_1.2,0.12165119647979736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,8,1,power_law_1.2,0.12639360427856444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,8,1,power_law_1.2,0.1324735999107361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,8,1,power_law_1.2,0.13895039558410643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,8,1,power_law_1.2,0.15400960445404052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,8,1,power_law_1.2,0.16618880033493041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,8,1,power_law_1.2,0.1793727993965149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,8,1,power_law_1.2,0.20149760246276854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,8,1,power_law_1.01,5.228070449829102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,8,1,power_law_1.2,0.21169281005859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,8,1,power_law_1.2,0.06087039709091187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,8,1,power_law_1.2,0.2407680034637451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,8,1,power_law_1.01,5.561209487915039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,8,1,power_law_1.2,0.27312641143798827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,8,1,power_law_1.2,0.35101439952850344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,8,1,power_law_1.2,0.06960639953613282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,8,1,power_law_1.2,0.07504640221595764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,8,1,power_law_1.2,0.09477120041847228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,8,1,power_law_1.2,0.09509119987487794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,8,1,power_law_1.2,0.09694079756736755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,8,1,power_law_1.2,0.09854080080986023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,8,1,power_law_1.2,0.09887999892234803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,8,1,power_law_1.2,0.10147839784622192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,8,1,power_law_1.2,0.1018496036529541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,8,1,power_law_1.2,0.4222911834716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,8,1,power_law_1.2,0.10900479555130005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,8,1,power_law_1.2,0.13507200479507447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,1,8,power_law_1.2,3.289798355102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,8,balanced,0.09178133805592854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,8,balanced,0.10313066840171814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,8,balanced,0.12360533078511556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,8,balanced,0.18553600708643594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,8,balanced,0.30800533294677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,8,1,power_law_1.2,0.02908160090446472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,8,balanced,0.5399200121561686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,8,1,power_law_1.2,0.03255679905414581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,8,balanced,0.7197973728179932
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,8,1,power_law_1.2,0.03924480080604553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,8,balanced,0.710752010345459
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,8,1,power_law_1.2,0.051667201519012454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,8,1,power_law_1.2,0.5806335926055908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,8,balanced,0.7045653661092123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,8,balanced,0.7090933322906494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,8,1,power_law_1.2,0.13824000358581542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,8,1,power_law_1.2,0.737830400466919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,8,1,power_law_1.2,0.1529728055000305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,8,1,power_law_1.2,0.1673151969909668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,8,1,power_law_1.2,0.17237119674682616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,8,1,power_law_1.2,0.05226879715919495
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,8,1,power_law_1.2,0.055078399181365964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,8,1,power_law_1.2,0.0588096022605896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,8,balanced,0.7409226894378662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,8,1,power_law_1.2,0.06412799954414368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,8,balanced,0.7326933542887369
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,8,1,power_law_1.2,0.060838401317596436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,8,balanced,0.7312959829966227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,8,1,power_law_1.2,0.06247680187225342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,8,balanced,0.756165345509847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,8,1,power_law_1.2,0.06793599724769592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,8,balanced,0.7645760377248129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,8,balanced,0.7921226819356283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,8,1,power_law_1.2,0.06897280216217042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,8,balanced,0.8200373649597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,8,1,power_law_1.2,0.07017599940299987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,8,balanced,0.8797653516133627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,8,1,power_law_1.2,0.20729598999023438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,8,1,power_law_1.2,0.07934079766273498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,8,balanced,0.8887893358866373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,8,1,power_law_1.2,0.07723519802093506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,8,balanced,1.0043893655141194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,8,1,power_law_1.2,0.08098559975624084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,8,1,power_law_1.2,2.725753593444824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,8,balanced,1.1625280380249023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,8,balanced,1.6065866152445476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,8,1,power_law_1.2,0.09193599820137024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,8,balanced,1.8937333424886067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,8,1,power_law_1.2,0.25761919021606444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,8,1,power_law_1.2,0.10683519840240478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,8,balanced,2.7086718877156577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,8,1,power_law_1.2,1.0539648056030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,8,1,power_law_1.2,0.3066368103027344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,8,balanced,3.5421387354532876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,8,1,power_law_1.2,0.12291200160980224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,8,balanced,5.7241973876953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,8,1,power_law_1.01,1.1113151550292968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,8,balanced,0.047925333182017006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,8,balanced,0.06610133250554402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,8,balanced,0.09431466460227966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,8,balanced,0.13181333740552267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,8,1,power_law_1.2,0.4305088043212891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,8,balanced,0.20502400398254395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,8,balanced,0.35134931405385333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,8,balanced,0.46898667017618817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,8,balanced,0.4747200012207031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,8,balanced,8.51522127787272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,8,balanced,0.4785439968109131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,8,balanced,0.4834666649500529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,8,balanced,0.4936853249867757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,8,balanced,0.5043466488520304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,8,balanced,0.5131733417510986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,8,balanced,0.060005332032839455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,8,balanced,0.5336906512578329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,8,balanced,0.06834133466084798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,8,balanced,0.5525493224461874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,8,1,power_law_1.2,0.15561599731445314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,8,balanced,0.5724639892578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,8,balanced,0.608789324760437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,8,balanced,0.684666633605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,8,balanced,0.7565600077311198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,8,1,power_law_1.2,0.5541056156158447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,8,balanced,0.921066681543986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,8,balanced,1.0738399823506672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,8,balanced,0.084714670976003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,8,1,power_law_1.2,1.3870783805847169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,8,balanced,0.11036800344785054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,8,balanced,0.16189866264661154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,8,balanced,0.27875200907389325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,8,balanced,0.28485333919525146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,8,balanced,0.2895359992980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,8,balanced,0.29230932394663495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,8,balanced,0.38369067509969074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,8,balanced,0.3932266632715861
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,8,1,power_law_1.2,0.1751039981842041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,8,balanced,0.40201600392659503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,8,balanced,0.41311466693878174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,8,balanced,0.4304800033569336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,8,balanced,0.44811733563741046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,8,balanced,0.4635839859644572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,8,balanced,0.4962879816691081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,8,balanced,0.5542506774266561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,8,balanced,0.5973333517710367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,8,balanced,0.7350719769795736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,8,balanced,0.8440586725870768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,8,balanced,1.5567413965861003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,8,balanced,1.2229866981506348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,8,balanced,1.8664906819661458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,8,balanced,2.6685546239217124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,1,8,power_law_1.2,13.10693817138672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,8,balanced,3.4639625549316406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,8,balanced,1.4500212669372559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,8,balanced,2.0917760531107583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,8,balanced,5.062149365743001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,8,balanced,2.739930788675944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,8,balanced,4.029013315836589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,8,balanced,6.813493092854817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,8,power_law_1.01,0.15228159427642823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,8,power_law_1.01,0.29938559532165526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,8,power_law_1.01,0.3615936040878296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,8,balanced,5.3180694580078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,8,power_law_1.01,0.36338560581207274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,8,power_law_1.01,0.116211199760437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,8,power_law_1.01,0.43912320137023925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,8,power_law_1.01,0.19765119552612304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,8,1,power_law_1.2,0.2554431915283203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,8,1,power_law_1.2,0.7804416179656982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,8,power_law_1.01,0.23514881134033203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,8,power_law_1.01,0.25767040252685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,1,8,power_law_1.2,34.47518615722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,8,balanced,0.029861333469549816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,8,balanced,0.03181866556406021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,8,balanced,0.03982399900754293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,8,power_law_1.01,0.27301759719848634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,8,balanced,0.061893333991368614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,8,balanced,0.08985066413879395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,8,balanced,0.14547200004259744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,8,power_law_1.01,0.31639039516448975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,8,balanced,0.14991999665896097
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,8,balanced,0.15375999609629312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,8,balanced,0.15201066931088766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,8,power_law_1.01,0.44022397994995116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,8,power_law_1.01,0.43802242279052733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,8,power_law_1.01,0.479366397857666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,8,power_law_1.01,0.4434624195098877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,8,power_law_1.01,0.6024127960205078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,8,power_law_1.01,0.45235838890075686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,8,power_law_1.01,0.5951935768127441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,8,power_law_1.01,0.455072021484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,8,power_law_1.01,0.6457151889801025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,8,power_law_1.01,0.4834559917449951
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,8,balanced,0.15307733416557312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,8,balanced,0.15492799878120422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,8,balanced,0.15440000096956888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,8,power_law_1.01,0.48659839630126955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,8,balanced,0.1606666644414266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,8,balanced,0.16402666767438254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,8,balanced,0.21313067277272543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,8,balanced,0.20485333601633707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,8,balanced,0.1904053290685018
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,8,balanced,0.36658668518066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,8,balanced,0.278714656829834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,8,balanced,0.2274613380432129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,8,balanced,0.24800533056259155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,8,balanced,0.28989867369333905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,8,balanced,0.3346560001373291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,8,power_law_1.01,0.5997951984405517
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,8,balanced,0.47778133551279706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,8,balanced,0.5663093328475952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,8,power_law_1.01,0.5142528057098389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,8,power_law_1.01,0.6274432182312012
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,8,balanced,0.81113068262736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,8,power_law_1.01,0.5346879959106445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,8,power_law_1.01,0.6247039794921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,8,balanced,1.060090700785319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,8,power_law_1.01,0.5797952175140381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,8,power_law_1.01,0.7045760154724121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,8,power_law_1.01,0.09415680170059204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,8,power_law_1.01,0.6410943984985351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,8,power_law_1.01,0.7236735820770264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,8,power_law_1.01,0.1584895968437195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,8,power_law_1.01,0.7590400218963623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,8,power_law_1.01,0.7367616176605225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,8,power_law_1.01,0.8702719688415528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,8,power_law_1.01,1.086246395111084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,8,power_law_1.01,1.3577407836914062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,8,power_law_1.01,1.7709823608398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,8,power_law_1.01,0.7584383964538575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,8,power_law_1.01,0.8048640251159668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,8,power_law_1.01,2.3218751907348634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,8,power_law_1.01,3.1791616439819337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,8,power_law_1.01,0.056415998935699464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,8,power_law_1.01,0.8653951644897461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,8,power_law_1.01,0.19276800155639648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,8,power_law_1.01,0.98221435546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,8,power_law_1.01,4.329344177246094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,8,power_law_1.01,1.17010555267334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,8,1,power_law_1.2,0.31496319770812986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,8,power_law_1.01,1.4001472473144532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,8,power_law_1.01,0.08433279991149903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,8,power_law_1.01,0.0978879988193512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,8,1,power_law_1.2,2.0196416854858397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,8,power_law_1.01,1.914975929260254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,8,power_law_1.01,0.10520319938659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,8,1,power_law_1.2,1.0137791633605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,8,power_law_1.01,6.265875244140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,8,power_law_1.01,0.11934720277786255
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,8,power_law_1.01,0.112716805934906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,8,power_law_1.01,0.1331007957458496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,8,power_law_1.01,0.1457919955253601
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,8,power_law_1.01,0.14085760116577148
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,8,power_law_1.01,0.12984960079193114
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,8,power_law_1.01,0.13630720376968383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,8,power_law_1.01,0.144377601146698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,8,power_law_1.01,2.3566207885742188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,8,power_law_1.01,0.15522559881210327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,8,power_law_1.01,8.314291381835938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,8,power_law_1.01,0.1521664023399353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,8,power_law_1.2,0.15313279628753662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,8,power_law_1.01,0.1647104024887085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,8,power_law_1.2,0.2929408073425293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,8,power_law_1.01,0.16799999475479127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,8,power_law_1.01,3.473388671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,8,power_law_1.2,0.2645888090133667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,8,power_law_1.01,0.20591359138488768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,8,power_law_1.2,0.35342719554901125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,8,power_law_1.01,0.2393280029296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,8,power_law_1.2,0.43330559730529783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,8,power_law_1.2,0.4328320026397705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,8,power_law_1.01,4.3780864715576175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,8,power_law_1.2,0.5892543792724609
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,8,power_law_1.01,0.17749119997024537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,8,power_law_1.2,0.5821311950683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,8,power_law_1.01,0.27470080852508544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,8,power_law_1.2,0.5834559917449951
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,8,power_law_1.01,0.23420801162719726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,8,power_law_1.01,0.227020788192749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,8,power_law_1.2,0.6089280128479004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,8,power_law_1.01,0.2443455934524536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,8,power_law_1.01,0.2554624080657959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,8,power_law_1.2,0.6040768146514892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,8,power_law_1.01,0.2640127897262573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,8,power_law_1.01,6.104723358154297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,8,power_law_1.2,0.6313920021057129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,8,power_law_1.2,0.6907008171081543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,8,1,power_law_1.2,1.4839872360229491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,8,power_law_1.2,0.7185984134674073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,8,power_law_1.2,0.751910400390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,8,power_law_1.2,0.7702271938323975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,8,power_law_1.01,0.33931519985198977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,8,power_law_1.2,0.825376033782959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,8,power_law_1.01,0.40767998695373536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,8,1,power_law_1.2,4.025017547607422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,8,power_law_1.01,8.502982330322265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,8,power_law_1.01,0.5665279865264893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,8,power_law_1.2,0.11819519996643066
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,8,power_law_1.01,0.6944831848144531
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,8,power_law_1.01,1.016153621673584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,8,power_law_1.2,0.8780735969543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,8,power_law_1.01,1.5771776199340821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,8,power_law_1.2,1.0138815879821776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,8,power_law_1.2,0.19879039525985717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,8,power_law_1.01,0.25386240482330324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,8,power_law_1.2,0.18449920415878296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,8,power_law_1.2,1.2663871765136718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,8,power_law_1.01,0.27599999904632566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,8,power_law_1.2,0.09550719857215881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,8,power_law_1.2,0.15779839754104613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,8,power_law_1.2,1.4923392295837403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,8,power_law_1.2,0.1629696011543274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,8,power_law_1.2,0.20111360549926757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,8,power_law_1.2,2.004960060119629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,8,power_law_1.2,0.25779199600219727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,8,power_law_1.2,0.23068161010742189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,8,power_law_1.2,0.25175039768218993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,8,power_law_1.2,2.5281471252441405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,8,power_law_1.2,0.2786880016326904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,8,power_law_1.2,0.3411072015762329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,8,power_law_1.2,0.45708160400390624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,8,power_law_1.2,0.430079984664917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,8,power_law_1.2,0.44208641052246095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,8,power_law_1.2,0.2582207918167114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,8,power_law_1.2,0.25335040092468264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,8,power_law_1.2,0.2726720094680786
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,8,1,power_law_1.2,0.4421247959136963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,8,power_law_1.2,0.35345919132232667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,8,power_law_1.2,0.363756799697876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,8,power_law_1.2,0.3725055932998657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,8,power_law_1.2,0.4509568214416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,8,power_law_1.2,0.38346879482269286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,8,power_law_1.2,0.4623104095458984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,8,power_law_1.2,0.40790400505065916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,8,power_law_1.2,0.4685952186584473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,8,power_law_1.2,0.4335807800292969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,8,power_law_1.2,0.4959616184234619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,8,power_law_1.2,3.7579647064208985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,8,power_law_1.2,0.45136637687683107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,8,power_law_1.2,0.5260223865509033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,8,power_law_1.2,0.5029183864593506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,8,power_law_1.2,0.559007978439331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,8,1,power_law_1.2,2.672115135192871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,8,power_law_1.2,0.5990975856781006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,8,power_law_1.01,0.34812800884246825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,8,power_law_1.2,4.480153656005859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,8,power_law_1.01,0.3696768045425415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,8,power_law_1.01,0.38776960372924807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,8,power_law_1.01,0.36522879600524905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,8,power_law_1.2,0.6064256191253662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,8,power_law_1.2,0.6534783840179443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,8,power_law_1.2,0.6817344188690185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,8,power_law_1.2,0.771347188949585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,8,power_law_1.2,7.299314880371094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,8,power_law_1.2,0.9189375877380371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,8,power_law_1.2,0.9004608154296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,8,power_law_1.2,1.1347968101501464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,8,power_law_1.2,1.1285375595092773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,8,power_law_1.2,1.4929535865783692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,8,power_law_1.2,1.8316287994384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,8,1,power_law_1.2,1.9381376266479493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,8,power_law_1.2,9.582182312011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,8,power_law_1.2,2.71409912109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,8,power_law_1.2,1.4216064453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,8,power_law_1.2,1.9880128860473634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,8,power_law_1.2,3.4683902740478514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,8,power_law_1.2,2.418604850769043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,8,power_law_1.2,3.337625503540039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,8,power_law_1.2,5.492089462280274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,8,power_law_1.01,0.4014400005340576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,8,power_law_1.01,0.407916784286499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,8,power_law_1.2,4.395321655273437
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,8,power_law_1.2,0.05625600218772888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,8,power_law_1.2,6.897798156738281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,8,power_law_1.2,0.08395519852638245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,8,1,balanced,0.06643733382225037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,8,1,balanced,0.0793333351612091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,8,1,balanced,0.09938666224479675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,8,1,balanced,0.14595199624697366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,8,1,balanced,0.31380265951156616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,8,1,balanced,0.5604319969813029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,8,power_law_1.01,0.4463935852050781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,8,1,balanced,0.5665760040283203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,8,power_law_1.2,6.112825775146485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,8,1,balanced,0.5565760135650635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,8,1,balanced,0.5606026649475098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,8,1,balanced,0.5607680082321167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,8,1,balanced,0.5622133413950602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,8,1,balanced,0.5621440013249716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,8,1,balanced,0.5641813278198242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,8,power_law_1.2,0.0833728015422821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,8,1,balanced,0.5679626862208048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,8,1,balanced,0.6734240055084229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,8,power_law_1.2,0.09461119771003723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,8,1,balanced,0.6820800304412842
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,8,power_law_1.2,0.12341760396957398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,8,1,balanced,0.7006026903788248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,8,1,balanced,0.7370346387227377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,8,1,balanced,0.7449653148651123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,8,1,balanced,0.7831359704335531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,8,1,balanced,0.8623200257619222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,8,1,balanced,1.2341333230336506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,8,power_law_1.2,0.11084159612655639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,8,1,balanced,1.3776480356852214
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,8,power_law_1.2,0.1435647964477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,8,1,balanced,1.9254186948140461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,8,1,power_law_1.2,0.5774847984313964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,8,power_law_1.2,0.12912000417709352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,8,power_law_1.2,0.12744319438934326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,8,1,balanced,2.4536959330240884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,8,1,balanced,3.613290786743164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,8,power_law_1.2,9.00799331665039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,8,1,balanced,0.041663999358812966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,8,1,balanced,0.07100266714890797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,8,1,balanced,0.09061866998672485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,8,1,balanced,0.12144000331560771
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,8,power_law_1.2,0.134879994392395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,8,1,balanced,0.22946133216222128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,8,power_law_1.2,0.13660800457000732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,8,1,balanced,5.398133595784505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,8,power_law_1.2,0.1346560001373291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,8,power_law_1.2,0.1503424048423767
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,8,power_law_1.2,0.15813759565353394
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,8,power_law_1.2,0.16142079830169678
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,8,power_law_1.2,0.1697343945503235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,8,power_law_1.2,0.1837440013885498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,8,1,balanced,0.39388267199198407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,8,power_law_1.2,0.2798271894454956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,8,1,balanced,0.39509332180023193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,8,1,balanced,0.3953813314437866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,8,power_law_1.2,0.2252608060836792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,8,1,balanced,0.39673598607381183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,8,power_law_1.01,0.4998335838317871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,8,1,balanced,0.4004266659418742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,8,power_law_1.2,0.25832960605621336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,8,1,balanced,0.40142401059468585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,8,power_law_1.01,0.5820159912109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,8,power_law_1.2,0.2971839904785156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,8,power_law_1.2,0.39151999950408933
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,8,power_law_1.2,0.444595193862915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,8,power_law_1.2,0.6285439968109131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,8,1,balanced,0.4090133508046468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,8,1,balanced,0.4127360184987386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,8,1,balanced,0.4198880195617676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,8,1,balanced,0.4676266511281331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,8,1,balanced,0.4795733292897542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,8,1,balanced,0.4986453453699748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,8,1,balanced,0.5468639930089315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,8,1,balanced,0.5851039886474609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,8,1,balanced,0.6595733165740967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,8,1,balanced,0.7525066534678141
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,8,power_law_1.2,0.8089728355407715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,8,1,balanced,1.108464002609253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,8,1,balanced,1.281717300415039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,8,power_law_1.2,1.1230591773986816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,8,1,balanced,1.8119680086771648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,8,power_law_1.2,1.8967744827270507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,8,1,balanced,2.3487146695454917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,8,1,power_law_1.2,5.520787048339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,8,1,power_law_1.01,0.06743040084838867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,8,1,power_law_1.01,0.07923200130462646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,8,1,power_law_1.01,0.09996799826622009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,8,1,power_law_1.01,0.14587520360946654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,8,1,power_law_1.01,0.21953918933868408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,8,1,power_law_1.01,0.3166016101837158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,8,1,power_law_1.01,0.4051072120666504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,8,1,balanced,3.400848070780436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,8,1,power_law_1.01,0.41472640037536623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,8,1,balanced,0.0473333348830541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,8,1,power_law_1.01,0.4209727764129639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,8,1,power_law_1.01,0.43724799156188965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,8,1,balanced,4.472954750061035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,8,1,power_law_1.01,0.4489408016204834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,8,1,power_law_1.01,0.040505599975585935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,8,1,power_law_1.01,0.494982385635376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,8,1,power_law_1.01,0.06399999856948853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,8,1,power_law_1.01,0.5051839828491211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,8,power_law_1.01,0.6439551830291748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,8,1,power_law_1.2,3.9349697113037108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,8,1,power_law_1.01,0.5457856178283691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,8,1,power_law_1.01,0.6017280101776123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,8,1,power_law_1.01,0.6632063865661622
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,8,1,balanced,0.02958933264017105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,8,1,balanced,0.03146666785081228
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,8,1,balanced,0.039706667264302574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,8,1,balanced,0.0621919979651769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,8,1,power_law_1.01,0.08826879858970642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,8,1,power_law_1.01,0.1205888032913208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,8,1,power_law_1.01,0.1958847999572754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,8,1,power_law_1.2,2.874086380004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,8,1,power_law_1.01,0.6843200206756592
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,8,1,balanced,0.09292266766230266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,8,1,balanced,0.15413332978884378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,8,1,balanced,0.15913599729537964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,8,1,balanced,0.16331199804941812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,8,1,balanced,0.16523733735084534
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,8,1,balanced,0.16739734013875326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,8,1,balanced,0.1723946730295817
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,8,1,balanced,0.17758933703104654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,8,1,balanced,0.18753600120544434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,8,1,power_law_1.01,0.23994240760803223
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,8,1,balanced,0.1943999926249186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,8,1,balanced,0.2216106653213501
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,8,1,balanced,0.23362666368484497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,8,1,power_law_1.01,0.30202879905700686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,8,1,balanced,0.2502613266309102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,8,1,balanced,0.25643734137217206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,8,1,power_law_1.01,0.33254399299621584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,8,1,power_law_1.01,0.7277184009552002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,8,1,power_law_1.01,0.33264639377593996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,8,1,power_law_1.01,0.775596809387207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,8,1,power_law_1.01,0.344812798500061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,8,1,power_law_1.01,0.35791358947753904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,8,1,power_law_1.01,0.8738112449645996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,8,1,power_law_1.01,0.38309760093688966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,8,1,power_law_1.01,1.019379234313965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,8,1,power_law_1.01,0.39943680763244627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,8,1,balanced,0.28143467505772907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,8,1,power_law_1.01,0.45776000022888186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,8,1,balanced,0.31484800577163696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,8,1,balanced,0.35277334849039715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,8,1,power_law_1.01,0.45637121200561526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,8,1,balanced,0.44197332859039307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,8,1,balanced,0.4999200105667114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,8,1,balanced,0.060138667623202004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,8,1,power_law_1.01,0.47655677795410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,8,1,balanced,0.757317304611206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,8,power_law_1.01,0.8654911994934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,8,1,power_law_1.01,0.5204031944274903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,8,1,balanced,0.7443626721700033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,8,1,power_law_1.01,0.5954112052917481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,8,power_law_1.01,1.0267840385437013
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,8,1,balanced,1.0538506507873535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,8,1,power_law_1.01,0.6425600051879883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,8,1,balanced,1.3757012685139973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,8,1,power_law_1.01,0.7699967861175537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,8,1,power_law_1.01,0.8897024154663086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,8,1,power_law_1.01,1.1548543930053712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,8,1,power_law_1.01,0.044588801264762876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,8,1,power_law_1.01,1.4106304168701171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,8,1,power_law_1.01,1.2680959701538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,8,1,power_law_1.01,0.05847679972648621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,8,1,power_law_1.01,0.08552320003509521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,8,1,power_law_1.01,1.5140671730041504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,8,1,power_law_1.01,1.927347183227539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,8,1,power_law_1.01,0.10849920511245728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,8,1,power_law_1.01,0.15219839811325073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,8,1,power_law_1.01,2.0379199981689453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,8,1,power_law_1.01,0.1931007981300354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,8,1,power_law_1.01,2.4519872665405273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,8,1,power_law_1.01,0.23888640403747557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,8,1,power_law_1.01,0.2483839988708496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,8,1,power_law_1.01,2.5668607711791993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,8,1,power_law_1.01,0.25470719337463377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,8,1,power_law_1.01,3.5057342529296873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,8,1,power_law_1.01,0.28494079113006593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,8,1,power_law_1.01,0.29485440254211426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,8,1,power_law_1.01,0.30332159996032715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,8,1,power_law_1.01,3.6507774353027345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,8,1,power_law_1.01,0.3173248052597046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,8,1,power_law_1.01,0.3355007886886597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,8,1,power_law_1.01,4.572102355957031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,8,1,power_law_1.01,0.3649215936660767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,8,1,power_law_1.01,0.3951551914215088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,8,1,power_law_1.01,0.41491198539733887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,8,1,power_law_1.01,4.677215957641602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,8,1,power_law_1.2,5.206003189086914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,8,1,power_law_1.01,0.5376575946807861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,8,1,power_law_1.2,0.06250879764556885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,8,1,power_law_1.2,0.04017280042171478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,8,1,power_law_1.2,0.07713279724121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,8,1,power_law_1.01,0.6390463829040527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,8,1,power_law_1.2,0.06988160014152527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,8,1,power_law_1.2,0.09805439710617066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,8,power_law_1.01,1.3967743873596192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,8,1,power_law_1.2,0.08762879967689514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,8,1,power_law_1.01,0.878502368927002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,8,1,balanced,0.08516800403594971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,8,1,power_law_1.2,0.14025599956512452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,8,1,power_law_1.2,0.11871999502182007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,8,1,power_law_1.2,0.20103039741516113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,8,1,power_law_1.01,1.1010175704956056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,8,1,power_law_1.2,0.16127359867095947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,8,1,power_law_1.2,0.29049599170684814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,8,1,power_law_1.2,0.228985595703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,8,1,power_law_1.2,0.40314879417419436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,8,1,power_law_1.2,0.29176959991455076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,8,1,power_law_1.01,2.390176010131836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,8,1,power_law_1.2,0.4116096019744873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,8,1,power_law_1.2,0.3266239881515503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,8,1,power_law_1.2,0.4173439979553223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,8,1,power_law_1.2,0.33552000522613523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,8,1,power_law_1.2,0.4230976104736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,8,1,power_law_1.2,0.334118390083313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,8,power_law_1.01,1.859071922302246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,8,1,power_law_1.01,2.944454383850098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,8,1,power_law_1.2,0.4703104019165039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,8,1,power_law_1.2,0.364518404006958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,8,1,power_law_1.01,0.029100799560546876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,8,1,power_law_1.2,0.4921152114868164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,8,1,power_law_1.01,0.031046399474143983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,8,1,power_law_1.2,0.5152128219604493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,8,1,power_law_1.01,0.04207360148429871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,8,power_law_1.01,2.5481216430664064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,8,1,power_law_1.2,0.5471615791320801
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,8,1,power_law_1.01,0.061459201574325564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,8,1,power_law_1.01,4.1777088165283205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,8,1,power_law_1.2,0.6396992206573486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,8,1,power_law_1.01,0.07979519963264466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,8,1,power_law_1.2,0.6709504127502441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,8,power_law_1.01,3.2858112335205076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,8,1,power_law_1.2,0.6897024154663086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,8,1,power_law_1.2,0.3828480005264282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,8,1,power_law_1.2,0.40705280303955077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,8,1,power_law_1.2,0.7393472194671631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,8,1,power_law_1.2,0.45733118057250977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,8,1,power_law_1.2,0.7791423797607422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,8,1,power_law_1.2,0.4695744037628174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,8,1,power_law_1.2,0.9337984085083008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,8,1,power_law_1.01,0.10865919589996338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,8,power_law_1.01,4.9311168670654295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,8,1,power_law_1.01,0.1341312050819397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,8,1,power_law_1.2,1.0233599662780761
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,8,1,power_law_1.01,0.14202239513397216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,8,1,power_law_1.01,0.14574079513549804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,8,1,power_law_1.2,1.2950207710266113
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,8,1,power_law_1.01,0.15083520412445067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,8,1,power_law_1.01,5.313792037963867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,8,1,power_law_1.01,0.15739519596099855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,8,1,power_law_1.2,1.5324735641479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,8,1,power_law_1.2,0.4961215972900391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,8,1,power_law_1.2,0.5481791973114014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,8,power_law_1.01,6.518899536132812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,8,1,power_law_1.2,0.5961664199829102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,8,1,power_law_1.2,0.6516223907470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,8,1,power_law_1.2,0.7790656089782715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,8,1,power_law_1.01,0.16631679534912108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,8,1,power_law_1.01,7.67589111328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,8,1,power_law_1.2,0.9004991531372071
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,8,1,power_law_1.01,0.1758784055709839
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,8,1,power_law_1.01,0.19037439823150634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,8,1,power_law_1.2,2.0580991744995116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,8,1,power_law_1.01,0.2084736108779907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,8,1,power_law_1.01,0.22456960678100585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,8,1,power_law_1.01,0.2600127935409546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,8,1,power_law_1.2,2.5857471466064452
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,8,1,power_law_1.01,0.25319039821624756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,8,1,balanced,0.11135466893513997
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,8,1,power_law_1.01,0.2979712009429932
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,8,1,power_law_1.01,0.31663360595703127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,8,1,power_law_1.2,1.1636159896850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,8,1,power_law_1.01,0.37706239223480226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,8,1,power_law_1.2,3.6414527893066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,8,1,power_law_1.01,9.999852752685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,8,1,power_law_1.01,0.4729663848876953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,8,1,power_law_1.01,0.6025728225708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,8,1,power_law_1.01,0.6445184230804444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,8,1,power_law_1.2,4.744480133056641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,8,balanced,0.04805333415667216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,8,balanced,0.08452799916267395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,8,balanced,0.097461332877477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,8,balanced,0.14122133453687033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,8,1,power_law_1.2,1.415116786956787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,8,balanced,0.2653333346048991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,8,balanced,0.46694934368133545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,8,balanced,0.5834293365478516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,8,balanced,0.6047093470891317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,8,1,power_law_1.2,1.936774444580078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,8,1,power_law_1.01,0.8095040321350098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,8,1,power_law_1.2,2.4602432250976562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,8,1,power_law_1.01,1.1245183944702148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,8,balanced,0.59988800684611
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,8,1,power_law_1.01,1.523084831237793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,8,balanced,0.5887733300526937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,8,1,power_law_1.2,0.028121599555015565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,8,balanced,0.045824001232783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,8,1,power_law_1.2,3.5164222717285156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,8,balanced,0.06192000210285187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,8,balanced,0.6212480068206787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,8,balanced,0.08649067083994548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,8,balanced,0.11946666240692139
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,8,1,power_law_1.2,0.030649599432945252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,8,balanced,0.6298186779022217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,8,balanced,0.18169599771499634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,8,balanced,0.3045813242594401
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,8,balanced,0.6367253462473551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,8,1,power_law_1.2,0.04338560104370117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,8,balanced,0.41278934478759766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,8,balanced,0.6071413358052572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,8,balanced,0.41658135255177814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,8,balanced,0.6542079846064249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,8,balanced,0.4213920036951701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,8,balanced,0.4251679976781209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,8,balanced,0.4361013174057007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,8,balanced,0.4464373191197713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,8,1,power_law_1.2,4.564057540893555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,8,balanced,0.45459731419881183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,8,balanced,0.4740159908930461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,8,balanced,0.4901119867960612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,8,balanced,0.5116106669108073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,8,balanced,0.05611733098824819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,8,balanced,0.5453973213831583
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,8,1,power_law_1.2,0.04089600145816803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,8,balanced,0.6149280071258545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,8,1,power_law_1.2,0.057548797130584715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,8,balanced,0.6842133204142252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,8,balanced,0.6698986689249674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,8,balanced,0.8222986857096354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,8,balanced,0.6970826784769694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,8,balanced,0.9633866945902506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,8,balanced,0.7296533584594727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,8,balanced,0.7571146488189697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,8,balanced,1.3907413482666016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,8,balanced,0.8856000105539957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,8,1,balanced,0.20957867304484049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,8,balanced,1.0302666823069255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,8,balanced,1.4369014104207356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,8,1,power_law_1.2,0.07530239820480347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,8,balanced,1.6954506238301594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,8,1,power_law_1.2,0.1016319990158081
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,8,1,power_law_1.2,0.1314687967300415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,8,balanced,2.415226618448893
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,8,1,power_law_1.2,0.1384511947631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,8,balanced,0.0587360014518102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,8,balanced,1.671024004618327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,8,balanced,0.07746666669845581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,8,balanced,3.1760266621907554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,8,balanced,2.3869706789652505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,8,balanced,3.0926027297973633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,8,balanced,5.081770579020183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,8,1,power_law_1.2,0.14391039609909057
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,8,1,power_law_1.2,0.15034879446029664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,8,1,power_law_1.2,0.15735679864883423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,8,balanced,4.515930811564128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,8,1,power_law_1.2,0.16611839532852174
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,8,1,power_law_1.2,0.17588479518890382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,8,1,power_law_1.2,0.1919360041618347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,8,balanced,6.89737065633138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,8,1,power_law_1.2,0.2080319881439209
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,8,1,power_law_1.2,0.23015038967132567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,8,1,power_law_1.2,0.2610111951828003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,8,balanced,6.1000213623046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,8,1,power_law_1.2,0.2605695962905884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,8,power_law_1.01,0.1296447992324829
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,8,1,power_law_1.2,0.30229759216308594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,8,power_law_1.01,0.2574527978897095
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,8,1,power_law_1.2,0.32200961112976073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,8,power_law_1.01,0.32369279861450195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,8,balanced,0.09605333209037781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,8,1,power_law_1.2,0.38329598903656004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,8,power_law_1.01,0.3237504005432129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,8,1,power_law_1.2,0.4724287986755371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,8,power_law_1.01,0.34936959743499757
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,8,1,power_law_1.2,0.5399360179901123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,8,power_law_1.01,0.38341760635375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,8,1,power_law_1.2,0.6551871776580811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,8,1,power_law_1.2,0.058873599767684935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,8,1,power_law_1.2,0.8273983955383301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,8,balanced,0.14325333635012308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,8,balanced,0.24075200160344443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,8,balanced,0.2462079922358195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,8,balanced,0.2482560078303019
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,8,1,power_law_1.2,1.135763168334961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,8,balanced,0.2536746660868327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,8,balanced,0.029711998999118805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,8,power_law_1.01,0.4833792209625244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,8,1,power_law_1.2,1.5666175842285157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,8,power_law_1.01,0.5001664161682129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,8,power_law_1.01,0.1061568021774292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,8,power_law_1.01,0.5383999824523926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,8,power_law_1.01,0.175654399394989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,8,power_law_1.01,0.20464639663696288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,8,power_law_1.01,0.24097280502319335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,8,power_law_1.01,0.24608640670776366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,8,balanced,0.03209600100914637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,8,balanced,0.037621334195137024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,8,balanced,0.05773866673310598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,8,power_law_1.01,0.23825280666351317
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,8,balanced,0.08216000099976857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,8,balanced,0.13000532984733582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,8,balanced,0.13201600313186646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,8,power_law_1.01,0.3972928047180176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,8,balanced,0.13778666655222574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,8,balanced,0.1344106694062551
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,8,balanced,0.13563733299573263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,8,power_law_1.01,0.3919167995452881
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,8,balanced,0.13822933038075766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,8,balanced,0.1379680037498474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,8,power_law_1.01,0.5294015884399415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,8,balanced,0.14036267002423605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,8,power_law_1.01,0.40159997940063474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,8,power_law_1.01,0.5137728214263916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,8,power_law_1.01,0.4046463966369629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,8,power_law_1.01,0.5148863792419434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,8,power_law_1.01,0.4133440017700195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,8,power_law_1.01,0.5532288074493408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,8,power_law_1.01,0.6070015907287598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,8,1,balanced,0.339408000310262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,8,power_law_1.01,0.6211071968078613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,8,1,power_law_1.2,0.8322239875793457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,8,balanced,0.14613866806030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,8,balanced,0.19514666001001993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,8,power_law_1.01,0.6102911949157714
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,8,balanced,0.18361600240071616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,8,balanced,0.1678239901860555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,8,balanced,0.3320320049921672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,8,power_law_1.01,0.6807551860809327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,8,balanced,0.22195200125376383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,8,power_law_1.01,0.42685441970825194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,8,balanced,0.19995200634002686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,8,power_law_1.01,0.43729281425476074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,8,balanced,0.21845867236455283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,8,balanced,0.25382934014002484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,8,power_law_1.01,0.45543680191040037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,8,balanced,0.2940586606661479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,8,1,balanced,0.3407306671142578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,8,balanced,0.42003198464711505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,8,power_law_1.01,0.48380160331726074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,8,balanced,0.49612800280253094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,8,power_law_1.01,0.5042816162109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,8,balanced,0.7083893616994222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,8,power_law_1.01,0.560748815536499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,8,balanced,0.33876800537109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,8,balanced,0.9229013125101725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,8,balanced,0.3489866654078166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,8,balanced,0.35810665289560956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,8,power_law_1.01,0.678111982345581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,8,power_law_1.01,0.7517568111419678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,8,balanced,0.36423468589782715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,8,power_law_1.01,0.7734848022460937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,8,power_law_1.01,0.8412735939025879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,8,power_law_1.01,0.9691519737243652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,8,power_law_1.01,1.0599424362182617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,8,power_law_1.01,1.1914431571960449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,8,1,power_law_1.2,0.08492799997329711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,8,power_law_1.01,1.2505536079406738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,8,power_law_1.01,1.6795583724975587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,8,power_law_1.01,1.6822912216186523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,8,power_law_1.01,2.0780799865722654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,8,power_law_1.01,2.0529279708862305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,8,power_law_1.01,2.822137641906738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,8,power_law_1.01,2.938105583190918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,8,power_law_1.01,3.6992256164550783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,8,power_law_1.01,3.7377857208251952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,8,power_law_1.01,0.0871616005897522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,8,power_law_1.01,5.712614440917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,8,power_law_1.01,5.7808574676513675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,8,balanced,0.38165334860483807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,8,balanced,0.3956693410873413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,8,balanced,0.4112906853357951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,8,power_law_1.01,7.05731201171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,8,power_law_1.01,7.291661071777344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,8,power_law_1.2,0.1297279953956604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,8,power_law_1.2,0.24920320510864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,8,power_law_1.2,0.22537600994110107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,8,power_law_1.2,0.32739839553833006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,8,power_law_1.2,0.38179199695587157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,8,power_law_1.01,0.14718719720840454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,8,1,power_law_1.2,3.8159870147705077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,8,power_law_1.2,0.3955199956893921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,8,power_law_1.2,0.10656640529632569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,8,power_law_1.2,0.503334379196167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,8,1,power_law_1.2,0.10689280033111573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,8,power_law_1.2,0.17456640005111695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,8,power_law_1.2,0.521017599105835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,8,power_law_1.2,0.17634559869766236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,8,power_law_1.2,0.5470784187316895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,8,1,balanced,0.34118398030598956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,8,1,balanced,0.3426719903945923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,8,power_law_1.01,0.05348479747772217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,8,power_law_1.01,0.07559679746627808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,8,power_law_1.2,0.18122880458831786
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,8,power_law_1.01,0.08743039965629577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,8,power_law_1.2,0.5082687854766845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,8,power_law_1.01,0.09926400184631348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,8,power_law_1.2,0.5267712116241455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,8,power_law_1.01,0.11695359945297241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,8,power_law_1.2,0.5729536056518555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,8,power_law_1.01,0.11943039894104004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,8,power_law_1.2,0.5468031883239746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,8,power_law_1.01,0.11818239688873292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,8,power_law_1.2,0.6138688087463379
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,8,power_law_1.01,0.1270848035812378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,8,power_law_1.01,0.11744639873504639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,8,power_law_1.2,0.6341567993164062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,8,power_law_1.2,0.2587968111038208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,8,power_law_1.01,0.11378560066223145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,8,balanced,0.4472373326619466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,8,power_law_1.2,0.6764351844787597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,8,power_law_1.2,0.27340800762176515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,8,balanced,0.5003786484400431
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,8,power_law_1.01,0.17763199806213378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,8,power_law_1.2,0.7029695987701416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,8,power_law_1.2,0.38334081172943113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,8,power_law_1.2,0.3921727895736694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,8,power_law_1.2,0.797056007385254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,8,power_law_1.2,0.39169280529022216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,8,power_law_1.2,0.8704256057739258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,8,power_law_1.2,0.40785918235778806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,8,power_law_1.2,1.0956095695495605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,8,power_law_1.2,0.4109951972961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,8,power_law_1.01,0.12933759689331054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,8,power_law_1.2,0.4235392093658447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,8,power_law_1.01,0.1247871994972229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,8,power_law_1.2,1.2761152267456055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,8,power_law_1.2,0.4353600025177002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,8,power_law_1.2,0.46353921890258787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,8,power_law_1.2,1.8076480865478515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,8,power_law_1.2,0.503872013092041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,8,power_law_1.2,0.54269437789917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,8,power_law_1.2,0.5749631881713867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,8,power_law_1.2,2.2922496795654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,8,power_law_1.01,0.13319679498672485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,8,power_law_1.01,0.13480960130691527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,8,1,balanced,0.330949326356252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,8,power_law_1.01,0.15109119415283204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,8,1,balanced,0.33217066526412964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,8,1,balanced,0.3359733422597249
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,8,power_law_1.01,0.1516991972923279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,8,1,balanced,0.3438986539840698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,8,1,balanced,0.3511360088984172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,8,power_law_1.01,0.15364480018615723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,8,1,balanced,0.3606666723887126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,8,1,balanced,0.3693600098292033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,8,1,balanced,0.386405348777771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,8,1,balanced,0.5158186753590902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,8,power_law_1.2,0.688979196548462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,8,1,balanced,0.5373013416926066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,8,power_law_1.2,0.790067195892334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,8,1,balanced,0.7692320346832275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,8,1,balanced,1.0068906943003337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,8,power_law_1.2,1.0387264251708985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,8,1,power_law_1.2,0.13963520526885986
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,8,power_law_1.01,0.23633921146392822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,8,power_law_1.2,1.2409472465515137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,8,power_law_1.01,0.20062079429626464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,8,power_law_1.01,0.20941441059112548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,8,balanced,0.5419679880142212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,8,power_law_1.2,1.6950912475585938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,8,power_law_1.01,0.24094080924987793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,8,power_law_1.2,3.2335166931152344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,8,power_law_1.01,0.29012479782104494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,8,power_law_1.2,2.1446271896362306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,8,power_law_1.2,4.3025470733642575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,8,power_law_1.2,3.104787254333496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,8,power_law_1.01,0.37662079334259035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,8,power_law_1.01,0.4934591770172119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,8,power_law_1.2,3.957727813720703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,8,power_law_1.01,0.6401088237762451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,8,power_law_1.2,6.027769470214844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,8,power_law_1.01,0.8747072219848633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,8,power_law_1.01,1.4093055725097656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,8,power_law_1.01,0.1773311972618103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,8,power_law_1.2,5.694457626342773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,8,power_law_1.2,0.08622080087661743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,8,power_law_1.2,0.1335935950279236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,8,power_law_1.2,8.33026580810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,8,power_law_1.2,0.15397759675979614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,8,balanced,0.668165365854899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,8,power_law_1.2,7.986847686767578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,8,1,balanced,0.05810666580994924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,8,balanced,0.7674773534138998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,8,1,balanced,0.06892266869544983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,8,1,balanced,0.08733333150545756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,8,1,balanced,0.129013329744339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,8,1,balanced,0.26263999938964844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,8,1,balanced,0.4723893404006958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,8,1,balanced,0.4748053153355916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,8,1,balanced,0.4667840003967285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,8,power_law_1.01,0.22272000312805176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,8,1,balanced,0.47140800952911377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,8,1,balanced,0.46855465571085614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,8,power_law_1.01,0.19689600467681884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,8,power_law_1.01,0.2162559986114502
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,8,power_law_1.2,0.052883201837539674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,8,power_law_1.2,0.07637760043144226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,8,power_law_1.2,0.1687872052192688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,8,power_law_1.2,0.078847998380661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,8,power_law_1.2,0.21906559467315673
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,8,power_law_1.2,0.09493119716644287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,8,1,power_law_1.2,1.0917311668395997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,8,1,balanced,0.03931200007597605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,8,1,balanced,0.06484266618887584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,8,1,balanced,0.07893866797288258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,8,1,balanced,0.10748799641927083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,8,1,balanced,0.19909866650899252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,8,1,balanced,0.340992013613383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,8,1,balanced,0.34031466643015545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,8,1,balanced,0.3423733313878377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,8,1,balanced,0.34197866916656494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,8,1,balanced,0.34410667419433594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,8,1,balanced,0.35023999214172363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,8,1,balanced,0.3531200091044108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,8,1,balanced,0.35582931836446124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,8,1,balanced,0.3640426794687907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,8,1,balanced,0.40670398871103924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,8,power_law_1.2,0.21337599754333497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,8,1,balanced,0.41728532314300537
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,8,power_law_1.2,0.11493120193481446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,8,1,balanced,0.43435200055440265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,8,1,balanced,0.47898133595784503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,8,power_law_1.2,0.23537919521331788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,8,1,balanced,0.4781920115152995
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,8,power_law_1.2,0.10299520492553711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,8,1,balanced,0.48368533452351886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,8,power_law_1.2,0.2205440044403076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,8,1,balanced,0.4905279874801636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,8,power_law_1.2,0.12094719409942627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,8,1,balanced,0.5572479963302612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,8,power_law_1.2,0.2309056043624878
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,8,power_law_1.2,0.12104320526123047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,8,1,balanced,0.5959786574045817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,8,1,balanced,0.6009493271509806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,8,power_law_1.2,0.12277120351791382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,8,1,balanced,0.6021759907404581
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,8,power_law_1.2,0.11764479875564575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,8,1,balanced,0.6136533419291178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,8,balanced,1.113258679707845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,8,power_law_1.2,0.12718720436096193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,8,1,balanced,0.6792746384938558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,8,balanced,1.3229599793752034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,8,1,balanced,0.4735786517461141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,8,1,balanced,0.505791982014974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,8,1,balanced,0.5767999887466431
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,8,power_law_1.01,0.23143680095672609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,8,balanced,1.9053653081258137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,8,1,balanced,0.6534666617711385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,8,power_law_1.01,0.23374719619750978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,8,1,balanced,0.95852263768514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,8,balanced,2.484597365061442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,8,1,balanced,1.1103520393371582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,8,1,balanced,1.5669652620951335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,8,power_law_1.2,0.12110079526901245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,8,1,balanced,0.7488799889882406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,8,1,balanced,2.1185332934061685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,8,power_law_1.2,0.13104000091552734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,8,balanced,3.6408745447794595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,8,1,balanced,2.031578699747721
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,8,power_law_1.2,0.13523199558258056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,8,1,balanced,2.6171679496765137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,8,1,balanced,2.935311953226725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,8,balanced,4.799802780151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,8,1,balanced,3.9226452509562173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,8,1,power_law_1.2,0.1739967942237854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,8,1,balanced,3.8623838424682617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,8,1,balanced,1.0480693181355794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,8,1,balanced,1.189242680867513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,8,power_law_1.2,0.14673919677734376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,8,1,balanced,0.04196266829967499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,8,1,balanced,5.109973271687825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,8,1,balanced,0.054005334774653115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,8,1,balanced,1.6502505938212078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,8,1,balanced,0.07685866455237071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,8,1,balanced,0.09945600231488545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,8,1,balanced,0.18242132663726807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,8,1,balanced,2.1000372568766275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,8,1,power_law_1.2,0.2356031894683838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,8,1,balanced,3.040506680806478
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,8,power_law_1.2,0.14598400592803956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,8,1,balanced,7.545573552449544
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,8,power_law_1.2,0.16007039546966553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,8,power_law_1.2,0.30935039520263674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,8,power_law_1.2,0.2227776050567627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,8,1,balanced,0.2975359956423442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,8,1,balanced,0.2995520035425822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,8,power_law_1.2,0.3068543910980225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,8,power_law_1.2,0.2242432117462158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,8,1,balanced,0.30340800682703656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,8,power_law_1.2,0.3282495975494385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,8,1,balanced,0.30078933636347455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,8,power_law_1.2,0.22755200862884523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,8,1,balanced,0.28513065973917645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,8,1,balanced,0.02975466599067052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,8,1,balanced,0.03180266668399175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,8,1,balanced,0.03857066730658213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,8,1,balanced,0.058277333776156105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,8,1,balanced,0.08469866712888081
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,8,1,balanced,0.13691199819246927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,8,1,balanced,0.13833600282669067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,8,1,balanced,0.1439520021279653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,8,1,balanced,0.14541332920392355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,8,1,balanced,9.951082865397135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,8,1,balanced,3.989813486735026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,8,1,power_law_1.01,0.04380159974098206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,8,1,power_law_1.01,0.037222400307655334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,8,1,power_law_1.01,0.06812160015106201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,8,1,power_law_1.01,0.05159040093421936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,8,1,power_law_1.01,0.08610560297966004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,8,1,balanced,0.28917332490285236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,8,1,power_law_1.01,0.07841920256614685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,8,1,balanced,0.2937600016593933
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,8,power_law_1.2,0.2505215883255005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,8,1,power_law_1.01,0.12743680477142333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,8,1,power_law_1.01,0.10622080564498901
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,8,power_law_1.2,0.3315392017364502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,8,1,power_law_1.01,0.18732800483703613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,8,1,power_law_1.01,0.16073600053787232
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,8,1,balanced,0.14748266339302063
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,8,1,balanced,0.15363732973734537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,8,1,power_law_1.01,0.21338241100311278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,8,1,power_law_1.01,0.2871040105819702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,8,power_law_1.2,0.3928704023361206
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,8,1,balanced,0.15782933433850607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,8,1,balanced,0.16265599926312765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,8,1,balanced,0.1726133426030477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,8,1,power_law_1.01,0.35034239292144775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,8,1,balanced,0.1953173279762268
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,8,power_law_1.2,0.5326591968536377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,8,1,balanced,0.2035199999809265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,8,1,balanced,0.21967999140421549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,8,power_law_1.2,0.3551680088043213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,8,1,balanced,0.2241599957148234
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,8,power_law_1.2,0.6709695816040039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,8,1,balanced,0.2465546727180481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,8,1,balanced,0.2974826693534851
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,8,1,balanced,0.2734559973080953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,8,1,balanced,0.3049493432044983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,8,1,balanced,0.3059200048446655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,8,1,balanced,0.3918079932530721
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,8,power_law_1.2,1.0094976425170898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,8,1,balanced,0.4347626765569051
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,8,1,balanced,0.6969119707743326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,8,1,balanced,0.6481120189030966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,8,1,power_law_1.01,0.2627840042114258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,8,power_law_1.2,1.534182357788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,8,1,balanced,0.9120852947235107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,8,1,power_law_1.01,0.28691840171813965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,8,1,power_law_1.01,0.35726079940795896
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,8,1,balanced,1.1867093245188396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,8,1,power_law_1.01,0.2936255931854248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,8,1,power_law_1.01,0.3602303981781006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,8,1,balanced,0.3133280078570048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,8,1,balanced,0.3243093291918437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,8,1,balanced,0.3380426565806071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,8,1,balanced,0.44576001167297363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,8,1,balanced,0.4646506706873576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,8,1,power_law_1.01,0.03988479971885681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,8,1,balanced,0.6648533344268799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,8,1,power_law_1.01,0.05246719717979431
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,8,1,balanced,0.85481063524882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,8,1,power_law_1.01,0.0763647973537445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,8,1,balanced,1.8162612915039062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,8,1,power_law_1.01,0.09684479832649232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,8,1,power_law_1.01,0.30035839080810545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,8,1,power_law_1.01,0.36129920482635497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,8,1,power_law_1.01,0.13411200046539307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,8,1,power_law_1.01,0.31071360111236573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,8,1,balanced,2.2418346405029297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,8,1,power_law_1.01,0.39159679412841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,8,1,power_law_1.01,0.33530240058898925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,8,1,power_law_1.01,0.418943977355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,8,1,power_law_1.01,0.34376959800720214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,8,1,power_law_1.01,0.43566718101501467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,8,1,power_law_1.01,0.3948863983154297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,8,1,power_law_1.01,0.463321590423584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,8,1,power_law_1.01,0.3953471899032593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,8,1,power_law_1.01,0.5224192142486572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,8,1,power_law_1.01,0.40554242134094237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,8,power_law_1.01,0.2989311933517456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,8,1,power_law_1.01,0.02839039862155914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,8,1,power_law_1.01,0.44336638450622556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,8,power_law_1.01,0.31857919692993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,8,1,power_law_1.01,0.030144000053405763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,8,1,power_law_1.01,0.16619520187377929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,8,1,power_law_1.01,0.5109632015228271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,8,power_law_1.01,0.3326848030090332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,8,1,power_law_1.01,0.03735679984092712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,8,1,power_law_1.01,0.20882558822631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,8,1,power_law_1.01,0.5580416202545166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,8,1,power_law_1.01,0.055801600217819214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,8,1,balanced,3.365621248881022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,8,1,power_law_1.01,0.21655681133270263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,8,1,power_law_1.01,0.0725823998451233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,8,1,power_law_1.01,0.22522239685058593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,8,1,power_law_1.01,0.2463167905807495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,8,power_law_1.2,0.3611327886581421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,8,1,balanced,4.386704126993815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,8,1,power_law_1.01,0.25571839809417723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,8,1,power_law_1.01,0.2668224096298218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,8,1,power_law_1.01,0.6621823787689209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,8,1,balanced,6.468234380086263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,8,power_law_1.2,0.3836927890777588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,8,power_law_1.2,0.39653759002685546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,8,power_law_1.01,0.34074881076812746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,8,1,power_law_1.01,0.27200000286102294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,8,1,power_law_1.01,0.09802240133285522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,8,1,power_law_1.2,0.2458944082260132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,8,1,power_law_1.01,0.2947904109954834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,8,1,power_law_1.01,0.11715840101242066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,8,1,power_law_1.2,0.2514240026473999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,8,1,power_law_1.01,0.3190399885177612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,8,1,power_law_1.01,0.12440320253372192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,8,1,power_law_1.01,0.3438463926315308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,8,1,balanced,8.516042709350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,8,1,power_law_1.01,0.1297152042388916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,8,1,power_law_1.01,0.36031360626220704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,8,1,power_law_1.2,0.033907198905944826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,8,1,power_law_1.01,0.13292160034179687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,8,1,power_law_1.01,0.46211838722229004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,8,1,power_law_1.01,0.13971199989318847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,8,1,power_law_1.01,0.5493055820465088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,8,1,power_law_1.01,0.758847999572754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,8,1,power_law_1.01,0.7692736148834228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,8,1,power_law_1.01,0.9553919792175293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,8,1,power_law_1.01,0.9973183631896972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,8,1,power_law_1.2,0.0679744005203247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,8,1,power_law_1.01,1.2203583717346191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,8,1,power_law_1.2,0.08550400137901307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,8,1,power_law_1.01,0.14699519872665406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,8,1,power_law_1.2,0.1220479965209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,8,1,power_law_1.01,0.1543936014175415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,8,1,power_law_1.01,1.6700672149658202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,8,1,power_law_1.2,0.17141760587692262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,8,1,power_law_1.01,0.5466879844665528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,8,1,power_law_1.2,0.24977920055389405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,8,1,power_law_1.01,0.5919807910919189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,8,1,power_law_1.01,2.1225343704223634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,8,1,power_law_1.2,0.3434943914413452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,8,1,power_law_1.01,0.6249855995178223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,8,1,power_law_1.2,0.3530879974365234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,8,1,power_law_1.01,0.6661952018737793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,8,1,power_law_1.2,0.3595263957977295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,8,1,power_law_1.01,0.7649663925170899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,8,1,power_law_1.01,3.0301952362060547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,8,1,power_law_1.01,0.16748160123825073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,8,1,power_law_1.01,0.8745535850524903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,8,1,power_law_1.01,0.18176000118255614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,8,1,power_law_1.01,0.19575680494308473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,8,1,power_law_1.01,1.090719985961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,8,1,power_law_1.01,0.22584960460662842
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,8,1,power_law_1.01,0.22033278942108153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,8,1,power_law_1.01,2.051545524597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,8,1,power_law_1.01,3.9442623138427733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,8,1,power_law_1.01,1.3085696220397949
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,8,1,power_law_1.01,0.25783040523529055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,8,1,power_law_1.2,0.0371071994304657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,8,1,power_law_1.2,0.3631743907928467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,8,power_law_1.2,0.44255361557006834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,8,1,power_law_1.2,0.048467200994491574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,8,1,power_law_1.01,1.7587648391723634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,8,1,power_law_1.01,2.547955131530762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,8,power_law_1.2,0.5296127796173096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,8,1,power_law_1.01,2.2167360305786135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,8,1,power_law_1.01,3.578073501586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,8,1,power_law_1.01,0.2718271970748901
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,8,1,power_law_1.01,0.32673280239105223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,8,1,power_law_1.2,0.3957567930221558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,8,1,power_law_1.2,0.07998719811439514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,8,1,power_law_1.01,3.143078422546387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,8,1,power_law_1.01,0.4028736114501953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,8,1,power_law_1.2,0.4207295894622803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,8,1,power_law_1.2,0.10499839782714844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,8,1,power_law_1.01,0.4622079849243164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,8,1,power_law_1.2,0.14160640239715577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,8,1,power_law_1.2,0.2006079912185669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,8,1,power_law_1.01,4.560076904296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,8,1,power_law_1.2,0.264902400970459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,8,1,power_law_1.01,4.06376953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,8,1,power_law_1.2,0.28072319030761717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,8,1,power_law_1.2,0.2814719915390015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,8,1,power_law_1.2,0.2902976036071777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,8,1,power_law_1.2,0.2943552017211914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,8,1,power_law_1.2,0.2941567897796631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,8,1,power_law_1.2,0.3041919946670532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,8,1,power_law_1.2,0.4432831764221191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,8,1,power_law_1.2,0.31640961170196535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,8,1,power_law_1.2,0.3179008007049561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,8,1,power_law_1.2,0.47820158004760743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,8,1,power_law_1.2,0.33680000305175783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,8,1,power_law_1.2,0.33815679550170896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,8,1,power_law_1.01,6.5897216796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,8,1,power_law_1.2,0.5628543853759765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,8,1,power_law_1.2,0.3562175989151001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,8,1,power_law_1.2,0.3714047908782959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,8,1,power_law_1.2,0.5704895973205566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,8,1,power_law_1.2,0.40250239372253416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,8,1,power_law_1.2,0.42143998146057127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,8,1,power_law_1.2,0.599616003036499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,8,1,power_law_1.2,0.5458559989929199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,8,1,power_law_1.2,0.6294464111328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,8,1,power_law_1.2,0.667903995513916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,8,1,power_law_1.01,0.5645823955535889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,8,1,power_law_1.2,0.8854720115661621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,8,1,power_law_1.01,0.7032832145690918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,8,1,power_law_1.2,0.40248961448669435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,8,power_law_1.01,0.33932158946990965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,8,1,power_law_1.2,0.4040192127227783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,8,1,power_law_1.2,1.1297151565551757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,8,power_law_1.01,0.37334399223327636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,8,1,power_law_1.01,0.9714176177978515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,8,1,power_law_1.01,8.579769897460938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,8,1,power_law_1.2,0.42135038375854494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,8,power_law_1.01,0.3934848070144653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,8,1,power_law_1.2,0.4759552001953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,8,1,power_law_1.01,1.3434687614440919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,8,power_law_1.2,0.6184959888458252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,8,balanced,0.02735466758410136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,8,1,power_law_1.2,0.5199295997619628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,8,1,power_law_1.2,0.5623295783996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,8,1,power_law_1.2,0.6786240100860595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,8,1,power_law_1.2,0.039315199851989745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,8,1,power_law_1.2,0.7780992031097412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,8,1,power_law_1.2,0.6704192161560059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,8,1,power_law_1.2,1.0122495651245118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,8,balanced,0.02757333219051361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,8,balanced,0.03137599925200144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,8,balanced,0.042352000872294106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,8,balanced,0.05801066756248474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,8,1,power_law_1.2,0.7797952175140381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,8,1,power_law_1.2,1.232249641418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,8,1,power_law_1.2,0.8816896438598633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,8,1,power_law_1.2,1.6827327728271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,8,1,power_law_1.2,1.1007679939270019
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,8,1,power_law_1.2,0.028070399165153505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,8,1,power_law_1.2,0.02999039888381958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,8,1,power_law_1.2,0.038438400626182555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,8,1,power_law_1.2,0.054790401458740236
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,8,1,power_law_1.2,0.07097600102424621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,8,1,power_law_1.2,2.1379520416259767
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,8,1,power_law_1.2,0.09507200121879578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,8,1,power_law_1.2,1.3284992218017577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,8,balanced,0.08078933258851369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,8,balanced,0.08240533371766408
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,8,1,power_law_1.2,0.11612800359725953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,8,balanced,0.08281599978605907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,8,balanced,0.08562666177749634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,8,balanced,0.08764266967773438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,8,1,power_law_1.2,0.12448639869689941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,8,balanced,0.08895466725031535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,8,balanced,0.09222933650016785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,8,1,power_law_1.2,1.7694656372070312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,8,balanced,0.09454400340716045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,8,1,power_law_1.2,0.12947200536727904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,8,balanced,0.09743466973304749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,8,1,power_law_1.2,3.051590347290039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,8,1,power_law_1.2,0.13180160522460938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,8,1,power_law_1.2,0.1395583987236023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,8,1,power_law_1.2,2.2297088623046877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,8,1,power_law_1.2,0.1469823956489563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,8,1,power_law_1.2,0.15593600273132324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,8,1,power_law_1.2,0.16981760263442994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,8,1,power_law_1.2,0.05292159914970398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,8,1,power_law_1.2,3.9553150177001952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,8,1,power_law_1.2,0.1845247983932495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,8,power_law_1.2,0.8106752395629883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,8,1,power_law_1.2,0.20035200119018554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,8,balanced,0.10367467006047566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,8,balanced,0.10950932900110881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,8,balanced,0.11755733688672383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,8,balanced,0.13522133231163025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,8,power_law_1.2,0.9842752456665039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,8,balanced,0.1511679987112681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,8,balanced,0.19933867454528809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,8,balanced,0.24297600984573364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,8,balanced,0.34911465644836426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,8,balanced,0.44068801403045654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,8,balanced,0.6340906620025635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,8,balanced,0.8215040365854899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,8,balanced,0.03141866624355316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,8,balanced,0.03182933231194814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,8,balanced,0.03364266703526179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,8,1,power_law_1.2,3.1663808822631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,8,balanced,1.2018933296203613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,8,1,power_law_1.2,0.2302464008331299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,8,balanced,1.5837759971618652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,8,1,power_law_1.2,0.22527360916137695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,8,balanced,0.03775466730197271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,8,balanced,0.04126933217048645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,8,balanced,0.04505600035190582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,8,balanced,0.04541333516438802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,8,balanced,0.05799466868241628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,8,balanced,0.05827199916044871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,8,balanced,0.06148266792297363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,8,balanced,0.06204266846179962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,8,balanced,0.06423999865849812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,8,balanced,0.066170667608579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,8,balanced,0.06855466465155284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,8,balanced,0.06845333178838094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,8,balanced,0.07334400216738383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,8,balanced,0.0376800000667572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,8,balanced,0.0479360024134318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,8,balanced,0.065610667069753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,8,balanced,0.06730666756629944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,8,balanced,0.07042666773001353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,8,balanced,0.07225599884986877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,8,balanced,0.07349333167076111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,8,balanced,0.07547733187675476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,8,balanced,0.07923733194669087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,8,balanced,0.08269866804281871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,8,balanced,0.08772266904513042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,8,1,power_law_1.2,0.2628351926803589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,8,balanced,0.09307733178138733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,8,balanced,0.10110933581988017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,8,balanced,0.11070400476455688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,8,1,power_law_1.2,0.07621759772300721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,8,balanced,0.13377066453297934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,8,1,power_law_1.2,0.28095359802246095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,8,balanced,0.15451733271280924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,8,balanced,0.21057599782943726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,8,balanced,0.2606079975763957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,8,balanced,0.36800531546274823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,8,balanced,0.07709866762161255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,8,balanced,0.0865226686000824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,8,balanced,0.09277866284052531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,8,balanced,0.1053493320941925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,8,balanced,0.12333333492279053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,8,balanced,0.14307733376820883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,8,balanced,0.19732266664505005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,8,balanced,0.24119466543197632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,8,balanced,0.3373440106709798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,8,balanced,0.4288533528645833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,8,power_law_1.01,0.441107177734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,8,balanced,0.6186346610387167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,8,1,power_law_1.2,4.077900695800781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,8,1,power_law_1.2,0.33245439529418946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,8,balanced,0.8083199659983317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,8,power_law_1.01,0.517574405670166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,8,balanced,0.4694026708602905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,8,1,power_law_1.2,0.4103360176086426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,8,balanced,1.1795307000478108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,8,balanced,0.6712906360626221
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,8,1,power_law_1.2,0.46961278915405275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,8,balanced,0.8737653096516927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,8,balanced,1.5540159543355305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,8,1,power_law_1.2,0.5721983909606934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,8,power_law_1.01,0.032416000962257385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,8,power_law_1.01,0.04251520037651062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,8,power_law_1.01,0.050380802154541014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,8,power_law_1.2,1.36561279296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,8,power_law_1.01,0.05079039931297302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,8,power_law_1.01,0.04987519979476929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,8,power_law_1.01,0.07740799784660339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,8,power_law_1.01,0.08014079928398132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,8,power_law_1.01,0.08095999956130981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,8,power_law_1.01,0.08190720081329346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,8,balanced,1.2819626331329346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,8,power_law_1.01,0.08462079763412475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,8,power_law_1.01,0.08672639727592468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,8,1,power_law_1.2,2.3855552673339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,8,1,power_law_1.2,0.09537919759750366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,8,balanced,1.68559996287028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,8,power_law_1.01,0.09049599766731262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,8,power_law_1.01,0.09174399971961975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,8,power_law_1.01,0.032620799541473386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,8,power_law_1.01,0.096806401014328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,8,power_law_1.01,0.03761920034885406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,8,power_law_1.01,0.1021183967590332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,8,power_law_1.01,0.04133760035037994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,8,power_law_1.2,1.6714111328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,8,1,power_law_1.2,3.0460351943969726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,8,power_law_1.2,2.4842687606811524
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,8,balanced,0.02548266698916753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,8,balanced,0.02697066714366277
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,8,balanced,0.027615999182065327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,8,balanced,0.027450665831565857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,8,balanced,0.03033600002527237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,8,balanced,0.03173333406448364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,8,1,power_law_1.2,4.1797950744628904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,8,power_law_1.01,0.1086527943611145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,8,power_law_1.01,0.040908798575401306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,8,power_law_1.01,0.11835520267486573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,8,power_law_1.2,3.3269630432128907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,8,power_law_1.01,0.13901439905166627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,8,power_law_1.01,0.1615231990814209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,8,power_law_1.01,0.2270143985748291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,8,1,power_law_1.2,0.7119487762451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,8,power_law_1.01,0.2791615962982178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,8,1,power_law_1.2,5.360396957397461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,8,balanced,0.03368533402681351
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,8,1,power_law_1.2,0.9814208030700684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,8,balanced,0.03479466587305069
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,8,balanced,0.035743998984495796
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,8,balanced,0.03645866612593333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,8,balanced,0.035989334185918175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,8,power_law_1.2,4.9608001708984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,8,power_law_1.01,0.043968001008033754
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,8,1,power_law_1.2,1.3763839721679687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,8,power_law_1.01,0.06509439945220948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,8,1,power_law_1.2,0.12082560062408447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,8,power_law_1.01,0.06572160124778748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,8,power_law_1.01,0.037503999471664426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,8,power_law_1.01,0.06912639737129211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,8,power_law_1.01,0.04444800019264221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,8,power_law_1.01,0.07002239823341369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,8,power_law_1.01,0.04598399996757507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,8,power_law_1.01,0.06952959895133973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,8,power_law_1.01,0.3823744058609009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,8,power_law_1.01,0.048172798752784726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,8,power_law_1.01,0.07335039973258972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,8,power_law_1.01,0.05415679812431336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,8,power_law_1.01,0.4951935768127441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,8,1,power_law_1.2,7.680409240722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,8,power_law_1.01,0.054688000679016115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,8,power_law_1.01,0.0565887987613678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,8,power_law_1.01,0.7168128013610839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,8,power_law_1.01,0.059462398290634155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,8,power_law_1.01,0.060031998157501223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,8,power_law_1.01,0.9475775718688965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,8,power_law_1.01,0.061766397953033444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,8,power_law_1.01,0.06439039707183838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,8,power_law_1.01,0.06769919991493226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,8,power_law_1.01,1.397817611694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,8,power_law_1.01,0.06995199918746949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,8,power_law_1.01,0.07957119941711426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,8,balanced,0.045184001326560974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,8,power_law_1.01,0.07975040078163147
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,8,balanced,0.04354133208592733
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,8,balanced,0.037920000652472176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,8,power_law_1.01,0.089273601770401
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,8,balanced,0.0643039991458257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,8,power_law_1.01,1.7145599365234374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,8,balanced,0.0566240002711614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,8,power_law_1.01,0.09553279876708984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,8,power_law_1.01,0.1021888017654419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,8,power_law_1.01,0.11416959762573242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,8,power_law_1.01,0.14430079460144044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,8,1,power_law_1.2,10.003167724609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,8,power_law_1.01,0.17021440267562865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,8,power_law_1.01,0.23054718971252441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,8,power_law_1.01,0.07797759771347046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,8,power_law_1.01,0.29075839519500735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,8,power_law_1.01,0.0875328004360199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,8,power_law_1.2,6.327590560913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,8,power_law_1.01,0.4049983978271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,8,power_law_1.01,0.09432960152626038
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,8,power_law_1.01,0.02547839879989624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,8,balanced,0.05199466645717621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,8,power_law_1.01,0.10911359786987304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,8,balanced,0.05129066606362661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,8,power_law_1.01,0.5149375915527343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,8,balanced,0.051882664362589516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,8,balanced,0.05733866492907206
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,8,balanced,0.06241066753864288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,8,balanced,0.08506133159001668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,8,power_law_1.01,0.7491519927978516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,8,power_law_1.01,0.9725824356079101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,8,power_law_1.2,0.033542400598526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,8,power_law_1.2,0.0394239991903305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,8,power_law_1.2,0.04496000111103058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,8,power_law_1.01,1.4491328239440917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,8,power_law_1.2,0.046854400634765626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,8,1,power_law_1.2,0.1546880006790161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,8,power_law_1.2,0.0326335996389389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,8,power_law_1.01,0.6017536163330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,8,power_law_1.01,0.1345471978187561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,8,power_law_1.2,0.03478400111198425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,8,1,power_law_1.2,0.20459520816802979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,8,power_law_1.01,0.15872000455856322
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,8,balanced,0.10290666421254475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,8,balanced,0.1379093329111735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,8,power_law_1.01,0.2155456066131592
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,8,balanced,0.1729546586672465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,8,balanced,0.2418079972267151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,8,power_law_1.01,0.2681663990020752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,8,balanced,0.3166026671727498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,8,power_law_1.01,0.36583681106567384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,8,power_law_1.2,0.03722879886627197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,8,power_law_1.01,0.47643518447875977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,8,power_law_1.2,0.04207360148429871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,8,power_law_1.2,0.047193598747253415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,8,power_law_1.2,0.0458624005317688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,8,power_law_1.01,0.6773888111114502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,8,power_law_1.2,0.07568640112876893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,8,power_law_1.2,0.04748159945011139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,8,power_law_1.2,0.07910400032997131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,8,power_law_1.2,0.036236798763275145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,8,power_law_1.2,0.05162879824638367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,8,power_law_1.2,0.080595201253891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,8,power_law_1.01,0.8886143684387207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,8,power_law_1.2,0.04415999948978424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,8,power_law_1.2,0.05600000023841858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,8,power_law_1.2,0.08318719863891602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,8,power_law_1.2,0.045747199654579164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,8,power_law_1.2,0.05692160129547119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,8,power_law_1.2,0.08403840065002441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,8,power_law_1.2,0.06346240043640136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,8,power_law_1.2,0.05899519920349121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,8,power_law_1.01,1.3061951637268066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,8,power_law_1.2,0.08680319786071777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,8,power_law_1.2,0.06534399986267089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,8,power_law_1.2,0.06082559823989868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,8,power_law_1.2,0.08878080248832702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,8,power_law_1.2,0.06208639740943909
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,8,power_law_1.2,0.09171839952468872
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,8,power_law_1.01,0.027001601457595826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,8,power_law_1.2,0.09692800045013428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,8,power_law_1.01,1.7461887359619142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,8,power_law_1.01,0.026899200677871705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,8,power_law_1.01,1.8478080749511718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,8,power_law_1.01,0.027859199047088622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,8,balanced,0.035536001125971474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,8,balanced,0.041637333730856575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,8,balanced,0.06926399966080983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,8,power_law_1.01,0.027750399708747864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,8,balanced,0.08920533458391826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,8,balanced,0.1378933290640513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,8,balanced,0.15599999825159708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,8,balanced,0.15879467129707336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,8,balanced,0.16193067034085593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,8,balanced,0.16357866923014322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,8,balanced,0.16554133097330728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,8,balanced,0.16894932587941489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,8,power_law_1.2,0.06702079772949218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,8,balanced,0.1747093399365743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,8,balanced,0.18153067429860434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,8,power_law_1.2,0.06748160123825073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,8,balanced,0.18988800048828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,8,balanced,0.20015466213226318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,8,power_law_1.2,0.06938880085945129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,8,balanced,0.21038933595021567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,8,power_law_1.2,0.025312000513076784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,8,balanced,0.23412799835205078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,8,power_law_1.2,0.07192959785461425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,8,balanced,0.27896533409754437
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,8,power_law_1.2,0.027212798595428467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,8,balanced,0.3232373396555583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,8,power_law_1.2,0.07777919769287109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,8,balanced,0.4732160170873006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,8,power_law_1.2,0.02705279886722565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,8,power_law_1.2,0.08055679798126221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,8,balanced,0.5622186660766602
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,8,power_law_1.2,0.027526399493217467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,8,power_law_1.01,0.030124801397323608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,8,balanced,0.7951412995656332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,8,power_law_1.2,0.028281599283218384
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,8,power_law_1.2,0.03000960052013397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,8,power_law_1.2,0.03112959861755371
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,8,power_law_1.2,0.033062401413917544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,8,power_law_1.2,0.0651968002319336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,8,power_law_1.2,0.03408640027046204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,8,power_law_1.2,0.10154240131378174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,8,power_law_1.2,0.06787840127944947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,8,power_law_1.2,0.10833920240402221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,8,power_law_1.2,0.07130240201950074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,8,power_law_1.2,0.11783039569854736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,8,power_law_1.2,0.07925760149955749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,8,power_law_1.2,0.14201600551605226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,8,power_law_1.2,0.08640000224113464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,8,power_law_1.2,0.17340799570083618
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,8,power_law_1.01,0.03129599988460541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,8,power_law_1.2,0.2324160099029541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,8,power_law_1.2,0.034483200311660765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,8,1,power_law_1.2,0.21790719032287598
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,8,power_law_1.2,0.03590399920940399
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,8,power_law_1.2,0.03731839954853058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,8,power_law_1.2,0.08865919709205627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,8,power_law_1.2,0.03840000033378601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,8,power_law_1.2,0.09485440254211426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,8,power_law_1.2,0.0374208003282547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,8,power_law_1.2,0.2923583984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,8,balanced,1.0122346878051758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,8,power_law_1.2,0.05299199819564819
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,8,power_law_1.2,0.04924159944057464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,8,balanced,1.4738826751708984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,8,power_law_1.2,0.40807042121887205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,8,power_law_1.2,0.049004799127578734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,8,power_law_1.2,0.0549888014793396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,8,balanced,1.9281387329101562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,8,power_law_1.2,0.05674239993095398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,8,1,power_law_1.2,0.2201472043991089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,8,power_law_1.2,0.5031295776367187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,8,power_law_1.2,0.0670527994632721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,8,power_law_1.2,0.09715840220451355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,8,balanced,2.971263885498047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,8,power_law_1.2,0.08081279993057251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,8,power_law_1.2,0.10186879634857178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,8,power_law_1.01,0.03284479975700379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,8,power_law_1.2,0.11226240396499634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,8,power_law_1.2,0.11786240339279175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,8,power_law_1.01,0.03418880105018616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,8,power_law_1.2,0.1352255940437317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,8,power_law_1.2,0.14695039987564087
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,8,power_law_1.01,0.03461759984493255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,8,power_law_1.2,0.16447999477386474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,8,power_law_1.2,0.17729920148849487
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,8,power_law_1.01,0.03497599959373474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,8,power_law_1.2,0.21743359565734863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,8,power_law_1.01,0.03899520039558411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,8,power_law_1.2,0.23740160465240479
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,8,power_law_1.01,0.03820799887180328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,8,power_law_1.2,0.2969280004501343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,8,power_law_1.2,0.2774847984313965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,8,power_law_1.01,0.03671039938926697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,8,power_law_1.01,0.05449600219726562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,8,power_law_1.2,0.10147199630737305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,8,power_law_1.2,0.4097919940948486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,8,power_law_1.01,0.05120000243186951
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,8,power_law_1.01,0.047040000557899475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,8,power_law_1.01,0.052179199457168576
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,8,power_law_1.2,0.12412159442901612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,8,power_law_1.01,0.05506560206413269
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,8,power_law_1.01,0.06519039869308471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,8,balanced,4.07040532430013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,8,balanced,0.04176533222198486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,8,power_law_1.01,0.07656319737434387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,8,balanced,0.04197866717974345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,8,power_law_1.2,0.3811072111129761
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,8,power_law_1.01,0.1018623948097229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,8,power_law_1.01,0.11639039516448975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,8,power_law_1.01,0.16282880306243896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,8,power_law_1.01,0.7917888164520264
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,8,power_law_1.01,0.20264959335327148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,8,1,power_law_1.2,0.244268798828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,8,power_law_1.2,0.4848896026611328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,8,power_law_1.01,0.28476800918579104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,8,power_law_1.01,0.9350848197937012
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,8,power_law_1.01,0.37085440158843996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,8,balanced,0.060218666990598045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,8,balanced,0.055498664577802025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,8,balanced,0.07817066709200542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,8,balanced,0.05620799958705902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,8,balanced,0.11544000109036763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,8,balanced,0.06690133114655812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,8,balanced,0.13783466815948486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,8,balanced,0.14099199573198953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,8,balanced,0.14244266351064047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,8,balanced,0.1458399991194407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,8,balanced,0.1471733351548513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,8,balanced,0.15308800339698792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,8,balanced,0.15891733765602112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,8,balanced,0.16619732975959778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,8,balanced,0.17722133795420328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,8,balanced,0.19084266821543375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,8,balanced,0.20318933327992758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,8,1,power_law_1.2,0.2522239923477173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,8,balanced,0.22675732771555582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,8,balanced,0.2796799937884013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,8,1,power_law_1.2,0.2676991939544678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,8,power_law_1.2,0.5395967960357666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,8,balanced,0.08051733175913493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,8,balanced,0.12299199899037679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,8,balanced,0.125791996717453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,8,power_law_1.2,0.7623871803283692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,8,balanced,0.12681066989898682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,8,balanced,0.1296266714731852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,8,balanced,0.132341335217158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,8,balanced,0.1353600025177002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,8,balanced,0.13981866836547852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,8,balanced,0.14522666732470194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,8,balanced,0.15240533153216043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,8,balanced,0.16171733538309732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,8,balanced,0.17499200503031412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,8,balanced,0.18513067563374838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,8,balanced,0.20853332678476968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,8,balanced,0.2551893393198649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,8,balanced,0.30107732613881427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,8,balanced,0.42510398228963214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,8,balanced,0.5114506483078003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,8,power_law_1.2,0.7027775764465332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,8,balanced,0.7204586664835612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,8,balanced,0.9317119916280111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,8,balanced,1.3514240582784016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,8,balanced,0.32942400376001996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,8,balanced,0.4665653308232625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,8,1,power_law_1.2,0.2780735969543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,8,balanced,0.5653653144836426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,8,1,power_law_1.2,0.2967679977416992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,8,balanced,0.8038880030314127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,8,1,power_law_1.2,0.32488958835601806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,8,balanced,1.0404799779256184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,8,1,power_law_1.2,0.35267200469970705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,8,balanced,1.5095252990722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,8,1,power_law_1.2,0.36713600158691406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,8,1,power_law_1.2,0.4725759983062744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,8,balanced,1.9796640078226726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,8,1,power_law_1.2,0.577791976928711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,8,1,power_law_1.2,0.7692480087280273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,8,balanced,2.9208265940348306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,8,1,power_law_1.2,0.9699007987976074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,8,balanced,3.860527992248535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,8,balanced,1.7665972709655762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,8,balanced,2.6027414004007974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,8,power_law_1.2,0.17327359914779664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,8,balanced,3.43885866800944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,8,power_law_1.2,0.7626304149627685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,8,power_law_1.01,0.0815936028957367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,8,power_law_1.01,0.09462400078773499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,8,power_law_1.01,0.10483200550079345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,8,power_law_1.01,0.11643520593643189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,8,power_law_1.01,0.1094208002090454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,8,power_law_1.2,1.0405183792114259
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,8,balanced,0.029135999580224354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,8,power_law_1.01,0.15073920488357545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,8,power_law_1.01,0.15406080484390258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,8,power_law_1.01,0.155404794216156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,8,power_law_1.01,0.1609663963317871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,8,power_law_1.01,0.16298240423202515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,8,power_law_1.01,0.1675711989402771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,8,power_law_1.2,0.978707218170166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,8,power_law_1.01,0.17575039863586425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,8,power_law_1.01,0.18069759607315064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,8,balanced,0.02958933264017105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,8,balanced,0.030576000610987347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,8,balanced,0.033674667278925575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,8,power_law_1.01,0.20038399696350098
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,8,balanced,0.05666666726271311
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,8,balanced,0.0602453351020813
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,8,balanced,0.0613973339398702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,8,balanced,0.06414933502674103
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,8,balanced,0.06493866443634033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,8,balanced,0.06445866823196411
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,8,balanced,0.06452266871929169
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,8,balanced,0.08317866424719493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,8,balanced,0.08069866895675659
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,8,balanced,0.07035199801127116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,8,balanced,0.16962132851282755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,8,balanced,0.15133866667747498
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,8,balanced,0.09548266728719075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,8,balanced,0.08521599570910136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,8,balanced,0.08989866574605306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,8,balanced,0.10275200009346008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,8,balanced,0.11524266997973125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,8,power_law_1.01,1.2758848190307617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,8,balanced,0.1516586641470591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,8,power_law_1.01,0.21043200492858888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,8,balanced,0.1771679917971293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,8,balanced,0.2468000054359436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,8,power_law_1.01,0.23427200317382812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,8,balanced,0.3193066716194153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,8,power_law_1.01,0.30346240997314455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,8,power_law_1.01,1.5941632270812989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,8,power_law_1.01,0.3404031991958618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,8,power_law_1.01,0.4118080139160156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,8,power_law_1.01,0.5300479888916015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,8,power_law_1.01,0.6545919895172119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,8,1,power_law_1.2,2.065395164489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,8,power_law_1.01,0.9632320404052734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,8,balanced,0.4567306836446126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,8,balanced,0.5952746470769247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,8,power_law_1.01,1.1945343971252442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,8,power_law_1.01,0.07069439888000488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,8,1,power_law_1.2,2.6053504943847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,8,power_law_1.01,0.08556159734725952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,8,power_law_1.01,2.334547233581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,8,power_law_1.01,0.08829439878463745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,8,power_law_1.2,0.20647680759429932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,8,power_law_1.01,1.843756866455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,8,power_law_1.01,0.09918720126152039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,8,power_law_1.01,0.10017280578613282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,8,power_law_1.01,0.13463679552078248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,8,power_law_1.01,0.13870079517364503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,8,power_law_1.01,3.004755210876465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,8,power_law_1.01,2.3702976226806642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,8,power_law_1.01,0.140665602684021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,8,power_law_1.01,0.14423680305480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,8,power_law_1.01,0.14758399724960328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,8,power_law_1.01,0.15301120281219482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,8,power_law_1.01,0.1618880033493042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,8,power_law_1.2,0.9126591682434082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,8,power_law_1.01,3.4637569427490233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,8,power_law_1.01,0.17550079822540282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,8,power_law_1.01,0.1990272045135498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,8,power_law_1.01,4.441036987304687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,8,power_law_1.01,0.2196671962738037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,8,1,power_law_1.2,3.582131195068359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,8,power_law_1.01,0.23573119640350343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,8,power_law_1.01,0.2657023906707764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,8,power_law_1.2,1.454918384552002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,8,power_law_1.01,0.32524800300598145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,8,power_law_1.01,4.725459289550781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,8,power_law_1.01,0.06854400038719177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,8,1,power_law_1.2,4.574694442749023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,8,power_law_1.01,5.881267166137695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,8,power_law_1.01,0.3929152011871338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,8,power_law_1.01,0.524403190612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,8,power_law_1.01,0.08321279883384705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,8,power_law_1.01,0.68472318649292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,8,1,power_law_1.2,6.580537414550781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,8,power_law_1.01,0.08313599824905396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,8,power_law_1.01,0.10025600194931031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,8,power_law_1.01,0.9748736381530761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,8,power_law_1.01,0.10903680324554443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,8,power_law_1.01,0.11571199893951416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,8,power_law_1.01,1.2043583869934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,8,power_law_1.01,0.12377599477767945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,8,power_law_1.01,0.1267135977745056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,8,power_law_1.01,0.12858879566192627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,8,power_law_1.01,0.03187839984893799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,8,power_law_1.01,1.806559944152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,8,power_law_1.01,0.035750401020050046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,8,power_law_1.01,0.04322560131549835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,8,power_law_1.01,0.04653440117835998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,8,power_law_1.01,0.045535999536514285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,8,power_law_1.01,2.286892890930176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,8,1,power_law_1.2,8.59288330078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,8,power_law_1.01,0.05658239722251892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,8,power_law_1.01,0.05772799849510193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,8,power_law_1.2,0.08028799891471863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,8,power_law_1.01,0.05934079885482788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,8,power_law_1.2,0.08339840173721313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,8,power_law_1.01,0.06106240153312683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,8,power_law_1.2,0.09168000221252441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,8,power_law_1.01,0.13146239519119263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,8,power_law_1.01,0.06157439947128296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,8,power_law_1.2,0.09290879964828491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,8,power_law_1.01,0.13597439527511596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,8,power_law_1.01,3.624038314819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,8,power_law_1.01,0.06056960225105286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,8,power_law_1.2,0.11674239635467529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,8,power_law_1.01,0.14223359823226928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,8,power_law_1.01,0.07107840180397033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,8,power_law_1.2,0.14862719774246216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,8,power_law_1.01,0.14864640235900878
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,8,power_law_1.01,0.07092480063438415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,8,power_law_1.2,0.15269119739532472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,8,power_law_1.01,0.1611840009689331
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,8,power_law_1.01,0.06786559820175171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,8,power_law_1.2,0.1551424026489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,8,power_law_1.01,0.17789440155029296
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,8,power_law_1.01,0.11765120029449463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,8,power_law_1.2,0.15710079669952393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,8,power_law_1.01,0.1931839942932129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,8,power_law_1.01,0.11134719848632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,8,power_law_1.2,0.16200319528579712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,8,power_law_1.01,0.23136000633239745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,8,power_law_1.01,0.08743039965629577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,8,power_law_1.2,0.16691839694976807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,8,power_law_1.01,4.633881759643555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,8,power_law_1.01,0.29513599872589114
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,8,power_law_1.01,0.08855680227279664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,8,power_law_1.2,0.16947200298309326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,8,power_law_1.01,0.09466239809989929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,8,power_law_1.2,0.18792959451675414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,8,power_law_1.01,0.11246720552444459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,8,power_law_1.2,0.20787839889526366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,8,power_law_1.01,0.13160320520401
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,8,power_law_1.2,0.23356800079345702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,8,power_law_1.01,0.17521920204162597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,8,power_law_1.2,0.276364803314209
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,8,power_law_1.01,0.20776960849761963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,8,power_law_1.2,0.29292800426483157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,8,power_law_1.01,0.303654408454895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,8,power_law_1.2,0.3566720008850098
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,8,power_law_1.01,0.38239359855651855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,8,power_law_1.01,0.3468287944793701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,8,power_law_1.01,0.5926464080810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,8,power_law_1.01,0.46524801254272463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,8,power_law_1.01,0.592083215713501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,8,power_law_1.01,0.8297663688659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,8,power_law_1.2,0.07021440267562866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,8,power_law_1.2,0.07794560194015503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,8,power_law_1.01,1.1026687622070312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,8,power_law_1.2,0.09844480156898498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,8,power_law_1.2,0.09690240025520325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,8,power_law_1.2,0.10008959770202637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,8,power_law_1.01,1.5506815910339355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,8,power_law_1.2,0.13063679933547973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,8,power_law_1.01,0.8077119827270508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,8,power_law_1.2,0.13784960508346558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,8,power_law_1.2,0.1407487988471985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,8,power_law_1.01,2.1045312881469727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,8,power_law_1.2,0.4111040115356445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,8,power_law_1.2,0.5561408042907715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,8,power_law_1.2,0.7154560089111328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,8,power_law_1.01,3.231455993652344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,8,power_law_1.2,0.06700159907341004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,8,power_law_1.2,1.4805439949035644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,8,power_law_1.2,0.1447551965713501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,8,power_law_1.2,0.07015039920806884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,8,power_law_1.2,0.1488319993019104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,8,power_law_1.2,0.08335360288619995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,8,power_law_1.2,0.15446399450302123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,8,power_law_1.2,0.09711359739303589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,8,power_law_1.2,0.1603775978088379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,8,power_law_1.01,4.126118469238281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,8,power_law_1.2,0.1093824028968811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,8,power_law_1.2,0.1776576042175293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,8,power_law_1.2,0.12031999826431275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,8,power_law_1.2,0.21224958896636964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,8,power_law_1.2,0.12433919906616211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,8,power_law_1.2,0.2241663932800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,8,power_law_1.2,0.12341760396957398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,8,power_law_1.2,0.2424704074859619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,8,power_law_1.2,0.12896640300750734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,8,power_law_1.2,0.2712575912475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,8,power_law_1.2,1.0295616149902345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,8,power_law_1.2,0.13215359449386596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,8,power_law_1.2,0.33844480514526365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,8,power_law_1.2,0.13819520473480223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,8,power_law_1.2,2.010316848754883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,8,power_law_1.2,0.4356800079345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,8,power_law_1.2,0.14158079624176026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,8,power_law_1.2,1.3257856369018555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,8,power_law_1.2,0.15216000080108644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,8,power_law_1.2,0.5774271965026856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,8,power_law_1.2,0.16346240043640137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,8,power_law_1.2,0.18122880458831786
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,8,power_law_1.2,0.304966402053833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,8,power_law_1.2,0.20135040283203126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,8,power_law_1.2,0.7005055904388428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,8,power_law_1.2,0.2379456043243408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,8,power_law_1.2,0.29806718826293943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,8,power_law_1.2,0.033318400382995605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,8,1,balanced,0.02738133321205775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,8,1,balanced,0.04548799991607666
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,8,power_law_1.2,0.03433600068092346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,8,1,balanced,0.060138667623202004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,8,1,balanced,0.08658666412035625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,8,1,balanced,0.13658666610717773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,8,power_law_1.2,0.0395904004573822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,8,1,balanced,0.1399946709473928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,8,1,balanced,0.1400320033232371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,8,1,balanced,0.1397386689980825
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,8,power_law_1.2,0.04714879989624023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,8,1,balanced,0.1399893363316854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,8,1,balanced,0.14129599928855896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,8,1,balanced,0.14309866229693094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,8,1,balanced,0.1711039940516154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,8,1,balanced,0.1770026683807373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,8,1,balanced,0.18090667327245077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,8,1,balanced,0.18439465761184692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,8,1,balanced,0.18979734182357788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,8,1,balanced,0.20100800196329752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,8,1,balanced,0.22049599885940552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,8,1,balanced,0.2440426747004191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,8,power_law_1.2,0.369049596786499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,8,1,balanced,0.3422079881032308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,8,1,balanced,0.3861546516418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,8,power_law_1.2,0.47796478271484377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,8,1,balanced,0.5206506649653116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,8,power_law_1.2,1.075705623626709
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,8,power_law_1.2,0.04933759868144989
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,8,power_law_1.2,0.05630080103874206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,8,power_law_1.2,1.3321920394897462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,8,power_law_1.2,0.6464704036712646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,8,1,balanced,0.648197333017985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,8,1,balanced,0.9328052997589111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,8,1,balanced,1.209114631017049
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,8,power_law_1.2,0.057715201377868654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,8,power_law_1.2,0.06024320125579834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,8,1,balanced,1.7782026926676433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,8,power_law_1.2,1.94335994720459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,8,1,balanced,2.340400060017904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,8,1,balanced,0.03276266654332479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,8,1,balanced,0.03331200033426285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,8,1,balanced,0.05277333160241445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,8,1,balanced,0.07806399961312611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,8,1,balanced,0.12736533085505167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,8,1,balanced,0.12921067078908285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,8,1,balanced,0.1288373370965322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,8,1,balanced,0.1288320024808248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,8,1,balanced,0.1318666636943817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,8,1,balanced,0.13085333506266275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,8,1,balanced,0.13461333513259888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,8,1,balanced,0.1492800017197927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,8,power_law_1.2,0.060031998157501223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,8,power_law_1.2,1.961676788330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,8,power_law_1.2,0.061459201574325564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,8,power_law_1.2,0.061689597368240354
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,8,power_law_1.2,0.06901760101318359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,8,power_law_1.2,0.8939776420593262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,8,1,balanced,0.15546666582425436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,8,power_law_1.2,2.499769592285156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,8,1,balanced,0.16159466902414957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,8,1,balanced,0.16901334126790366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,8,1,balanced,0.1753013332684835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,8,1,balanced,0.18784532944361368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,8,1,balanced,0.21341866254806519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,8,1,balanced,0.24241065979003906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,8,1,balanced,0.3307253321011861
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,8,power_law_1.2,0.06830719709396363
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,8,power_law_1.2,0.06798080205917359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,8,power_law_1.2,0.11489280462265014
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,8,power_law_1.2,0.11240320205688477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,8,power_law_1.2,0.08670719861984252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,8,power_law_1.2,0.09336959719657897
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,8,power_law_1.2,0.09960960149765015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,8,power_law_1.2,0.12012159824371338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,8,1,balanced,0.38277868429819745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,8,power_law_1.2,0.13983360528945923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,8,1,balanced,0.5235199928283691
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,8,power_law_1.2,0.18507519960403443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,8,1,balanced,0.6655679941177368
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,8,power_law_1.2,0.23157119750976562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,8,1,balanced,0.9534933567047119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,8,1,balanced,1.23853333791097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,8,1,balanced,1.8296640714009602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,8,1,balanced,2.4124906857808432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,8,power_law_1.2,1.171788787841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,8,power_law_1.2,1.3782912254333497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,8,power_law_1.2,1.6280832290649414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,8,1,balanced,0.027461332579453785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,8,1,balanced,0.02922133356332779
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,8,1,balanced,0.030586667358875275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,8,1,balanced,0.03793599953254064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,8,1,balanced,0.060175999999046326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,8,1,balanced,0.06164266665776571
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,8,1,balanced,0.06472533444563548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,8,power_law_1.2,2.6946943283081053
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,8,1,balanced,0.06811200082302094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,8,1,balanced,0.0679253339767456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,8,1,balanced,0.0693280001481374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,8,1,balanced,0.07251733541488647
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,8,1,balanced,0.08434133728345235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,8,power_law_1.2,0.31221120357513427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,8,1,balanced,0.08575466275215149
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,8,1,balanced,0.09055999914805095
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,8,1,balanced,0.09083200494448344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,8,1,balanced,0.09286399682362874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,8,1,balanced,0.0983786682287852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,8,1,balanced,0.10897599657376607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,8,1,balanced,0.11795199910799663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,8,1,balanced,0.15813866257667542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,8,1,balanced,0.1727466583251953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,8,1,balanced,0.2878986597061157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,8,1,balanced,0.24555732806523642
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,8,1,balanced,0.34623467922210693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,8,power_law_1.2,2.225574493408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,8,1,balanced,0.44226666291554767
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,8,1,balanced,0.6276853481928507
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,8,1,balanced,0.817477305730184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,8,1,power_law_1.01,0.026240000128746034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,8,1,power_law_1.01,0.04615679979324341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,8,1,power_law_1.01,0.05842559933662415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,8,1,power_law_1.01,0.07400959730148315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,8,1,power_law_1.01,0.08965759873390197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,8,1,power_law_1.01,0.12121599912643433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,8,1,power_law_1.01,0.12554880380630493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,8,1,power_law_1.01,0.12718720436096193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,8,1,power_law_1.01,0.13373440504074097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,8,1,power_law_1.01,0.13679360151290892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,8,1,power_law_1.01,0.14625920057296754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,8,1,power_law_1.01,0.16381440162658692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,8,1,power_law_1.01,0.16860159635543823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,8,1,power_law_1.01,0.1751487970352173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,8,1,power_law_1.01,0.18040319681167602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,8,1,power_law_1.01,0.18652160167694093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,8,1,power_law_1.01,0.20138239860534668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,8,1,power_law_1.01,0.26759040355682373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,8,1,power_law_1.01,0.29434239864349365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,8,1,power_law_1.01,0.34742400646209715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,8,1,power_law_1.01,0.41710720062255857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,8,1,power_law_1.01,0.5540095806121826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,8,1,power_law_1.01,0.6976255893707275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,8,1,power_law_1.01,0.9757887840270996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,8,1,power_law_1.01,1.2548800468444825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,8,1,power_law_1.01,1.8236608505249023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,8,1,power_law_1.01,2.386739158630371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,8,1,power_law_1.01,0.030188798904418945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,8,1,power_law_1.01,0.044896000623703004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,8,power_law_1.2,3.3584320068359377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,8,1,power_law_1.01,0.052876800298690796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,8,1,power_law_1.01,0.06845440268516541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,8,1,power_law_1.01,0.0806335985660553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,8,1,power_law_1.01,0.10641920566558838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,8,1,power_law_1.01,0.11006720066070556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,8,1,power_law_1.01,0.110534405708313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,8,1,power_law_1.01,0.12186880111694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,8,1,power_law_1.01,0.13046400547027587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,8,1,power_law_1.01,0.14113919734954833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,8,1,power_law_1.01,0.14683519601821898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,8,1,power_law_1.01,0.15221760272979737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,8,1,power_law_1.01,0.16412800550460815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,8,1,power_law_1.01,0.17192959785461426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,8,1,power_law_1.01,0.18906879425048828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,8,1,power_law_1.01,0.20482559204101564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,8,1,power_law_1.01,0.244486403465271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,8,1,power_law_1.01,0.2764415979385376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,8,1,power_law_1.01,0.34554879665374755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,8,power_law_1.2,0.4044928073883057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,8,1,power_law_1.01,0.4168831825256348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,8,1,power_law_1.01,0.5579904079437256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,8,1,power_law_1.01,0.7043456077575684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,8,1,power_law_1.01,0.9881407737731933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,8,1,power_law_1.01,1.284928035736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,8,1,power_law_1.01,1.8621696472167968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,8,1,power_law_1.01,2.4490432739257812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,8,power_law_1.2,0.5805503845214843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,8,1,power_law_1.01,0.02595840096473694
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,8,1,power_law_1.01,0.02691200077533722
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,8,1,power_law_1.01,0.0302592009305954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,8,1,power_law_1.01,0.035622400045394895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,8,1,power_law_1.01,0.04454399943351746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,8,1,power_law_1.01,0.05678079724311828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,8,1,power_law_1.01,0.061350399255752565
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,8,1,power_law_1.01,0.06367999911308289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,8,1,power_law_1.01,0.06721919775009155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,8,1,power_law_1.01,0.06791679859161377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,8,1,power_law_1.01,0.07348480224609374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,8,1,power_law_1.01,0.07810559868812561
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,8,1,power_law_1.01,0.08435840010643006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,8,1,power_law_1.01,0.09221119880676269
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,8,1,power_law_1.01,0.0855296015739441
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,8,1,power_law_1.01,0.09068160057067871
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,8,1,power_law_1.01,0.10105600357055664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,8,1,power_law_1.01,0.10688639879226684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,8,1,power_law_1.01,0.1207808017730713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,8,1,power_law_1.01,0.16165759563446044
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,8,1,power_law_1.01,0.18504960536956788
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,8,1,power_law_1.01,0.21679360866546632
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,8,1,power_law_1.01,0.2821824073791504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,8,1,power_law_1.01,0.36037120819091795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,8,1,power_law_1.01,0.46189441680908205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,8,power_law_1.2,0.42369918823242186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,8,1,power_law_1.01,0.6454720020294189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,8,1,power_law_1.2,0.025567999482154845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,8,1,power_law_1.2,0.04624640047550201
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,8,1,power_law_1.01,0.8491392135620117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,8,1,power_law_1.2,0.057081598043441775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,8,1,power_law_1.2,0.030214399099349976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,8,1,power_law_1.2,0.0709119975566864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,8,1,power_law_1.2,0.04622080028057098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,8,1,power_law_1.2,0.08158079981803894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,8,1,power_law_1.2,0.052716797590255736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,8,1,power_law_1.2,0.11840640306472779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,8,1,power_law_1.2,0.06575359702110291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,8,1,power_law_1.2,0.12543359994888306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,8,1,power_law_1.2,0.07930880188941955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,8,1,power_law_1.2,0.13124480247497558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,8,1,power_law_1.2,0.106387197971344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,8,1,power_law_1.2,0.13585920333862306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,8,1,power_law_1.2,0.1081536054611206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,8,1,power_law_1.2,0.13845759630203247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,8,1,power_law_1.2,0.11130880117416382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,8,1,power_law_1.2,0.15770879983901978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,8,1,power_law_1.2,0.12976640462875366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,8,1,power_law_1.2,0.16508159637451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,8,1,power_law_1.2,0.13303040266036986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,8,1,power_law_1.2,0.170905601978302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,8,1,power_law_1.2,0.14062080383300782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,8,1,power_law_1.2,0.17568639516830445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,8,1,power_law_1.2,0.14962559938430786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,8,1,power_law_1.2,0.1814463973045349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,8,1,power_law_1.2,0.15770879983901978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,8,1,power_law_1.2,0.18934400081634523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,8,1,power_law_1.2,0.16528639793395997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,8,1,power_law_1.2,0.20434560775756835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,8,1,power_law_1.2,0.17531520128250122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,8,1,power_law_1.2,0.2577728033065796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,8,1,power_law_1.2,0.19144959449768068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,8,1,power_law_1.2,0.2944447994232178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,8,1,power_law_1.2,0.20617599487304689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,8,1,power_law_1.2,0.24507520198822022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,8,1,power_law_1.2,0.2779968023300171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,8,1,power_law_1.2,0.3501760005950928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,8,1,power_law_1.2,0.4203904151916504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,8,1,power_law_1.2,0.5661312103271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,8,1,power_law_1.2,0.34764800071716306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,8,1,power_law_1.2,0.4188864231109619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,8,power_law_1.2,3.8276737213134764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,8,power_law_1.2,3.8974014282226563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,8,1,power_law_1.2,0.5555263996124268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,8,1,power_law_1.2,0.6967743873596192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,8,1,power_law_1.2,0.7033599853515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,8,power_law_1.2,1.8035327911376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,8,1,power_law_1.2,0.9784000396728516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,8,1,power_law_1.2,0.025913599133491515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,8,1,power_law_1.2,0.02686080038547516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,8,1,power_law_1.2,0.030687999725341798
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,8,1,power_law_1.2,0.034457600116729735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,8,1,power_law_1.2,0.04328320026397705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,8,1,power_law_1.2,0.05797759890556335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,8,1,power_law_1.2,0.06223999857902527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,8,1,power_law_1.2,0.0650816023349762
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,8,1,power_law_1.2,0.06828160285949707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,8,1,power_law_1.2,0.06968320012092591
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,8,1,power_law_1.2,0.07437440156936645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,8,1,power_law_1.2,0.0788096010684967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,8,1,power_law_1.2,0.08401920199394226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,8,1,power_law_1.2,0.09189119935035706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,8,1,power_law_1.2,0.08570240139961242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,8,1,power_law_1.2,0.0911296010017395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,8,1,power_law_1.2,1.258060836791992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,8,1,power_law_1.2,0.9906304359436036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,8,power_law_1.2,1.826655960083008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,8,balanced,0.07622933387756348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,8,balanced,0.08853333195050557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,8,balanced,0.1055466632048289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,8,balanced,0.16209600369135538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,8,balanced,0.3239893317222595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,8,balanced,0.4288426637649536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,8,balanced,0.4322293202082316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,8,balanced,0.43481600284576416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,8,1,power_law_1.2,0.1017024040222168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,8,balanced,0.43988267580668133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,8,balanced,0.4423946539560954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,8,balanced,0.44919466972351074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,8,balanced,0.45877333482106525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,8,balanced,0.4646986722946167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,8,balanced,0.4801386594772339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,8,balanced,0.49084266026814777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,8,balanced,0.50709335009257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,8,balanced,0.5345973173777262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,8,balanced,0.590282678604126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,8,balanced,0.6573386589686075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,8,1,power_law_1.2,0.10703999996185302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,8,balanced,0.9289226531982422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,8,1,power_law_1.2,0.12303999662399293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,8,balanced,1.063637336095174
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,8,1,power_law_1.2,0.16491520404815674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,8,1,power_law_1.2,0.18747520446777344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,8,balanced,1.5106186866760254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,8,1,power_law_1.2,0.22064640522003173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,8,balanced,1.916261355082194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,8,balanced,2.6960585912068686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,8,balanced,3.7854134241739907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,8,1,power_law_1.2,0.29068799018859864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,8,balanced,5.934682846069336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,8,power_law_1.2,5.734598541259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,8,balanced,0.052202666799227394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,8,balanced,0.07382399837176006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,8,balanced,0.09210667014122009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,8,balanced,0.13776533802350363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,8,balanced,0.20298133293787637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,8,balanced,0.30586665868759155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,8,balanced,0.31031999985376996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,8,1,power_law_1.2,1.2889535903930665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,8,balanced,0.3158239920934041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,8,balanced,0.31860800584157306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,8,balanced,0.3244853417078654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,8,balanced,0.3346720139185588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,8,balanced,0.34388267993927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,8,balanced,0.35077865918477374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,8,balanced,8.09011713663737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,8,balanced,0.3717333475748698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,8,balanced,0.3911840120951335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,8,balanced,0.41073067982991535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,8,balanced,0.4458666642506917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,8,balanced,0.5185600121816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,8,balanced,0.5932586590449015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,8,balanced,0.8706560134887695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,8,balanced,1.0176586310068767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,8,balanced,1.445365269978841
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,8,1,power_law_1.2,0.3607680082321167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,8,balanced,1.8753172556559246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,8,power_law_1.2,4.756582260131836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,8,balanced,2.609018643697103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,8,balanced,3.4638665517171225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,8,balanced,5.102869351704915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,8,balanced,0.057189335425694786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,8,balanced,0.0724533349275589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,8,balanced,7.174336115519206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,8,power_law_1.01,0.15143040418624878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,8,power_law_1.01,0.3230655908584595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,8,power_law_1.01,0.25515520572662354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,8,power_law_1.01,0.3211456060409546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,8,balanced,0.03141333411137263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,8,balanced,0.03212266663710276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,8,balanced,0.045893331368764244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,8,power_law_1.01,0.33522560596466067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,8,balanced,0.0645066648721695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,8,balanced,0.09283199906349182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,8,balanced,0.1093280017375946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,8,power_law_1.01,0.4169600009918213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,8,balanced,0.11132267117500305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,8,1,power_law_1.2,1.8705984115600587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,8,balanced,0.114138662815094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,8,balanced,0.11342933773994446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,8,balanced,0.11382399996121724
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,8,balanced,0.11573333541552226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,8,balanced,0.11781866351763408
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,8,balanced,0.1731520096460978
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,8,balanced,0.15427733461062113
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,8,balanced,0.14246400197347006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,8,balanced,0.3334133227666219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,8,balanced,0.2606719930966695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,8,balanced,0.15432533621788025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,8,balanced,0.15938133001327515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,8,balanced,0.17622933785120645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,8,balanced,0.19593600432078043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,8,1,power_law_1.2,0.461568021774292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,8,balanced,0.27739200989405316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,8,power_law_1.01,0.4140160083770752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,8,balanced,0.31108800570170086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,8,power_law_1.01,0.4212672233581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,8,balanced,0.44137601057688397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,8,balanced,0.573578675587972
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,8,balanced,0.7796213626861572
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,8,balanced,1.1386826833089192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,8,power_law_1.01,0.12465920448303222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,8,power_law_1.01,0.19566719532012938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,8,power_law_1.01,0.18320640325546264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,8,balanced,0.0865760048230489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,8,power_law_1.01,0.4282112121582031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,8,1,power_law_1.2,1.8182655334472657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,8,power_law_1.01,0.20147840976715087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,8,power_law_1.01,0.43659520149230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,8,power_law_1.01,0.19349759817123413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,8,power_law_1.01,0.44594559669494627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,8,power_law_1.01,0.2959424018859863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,8,power_law_1.01,0.45036802291870115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,8,power_law_1.01,0.30005760192871095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,8,power_law_1.01,0.45768961906433103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,8,power_law_1.01,0.3036736011505127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,8,power_law_1.01,0.47875199317932127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,8,power_law_1.01,0.3112384080886841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,8,balanced,0.10459199547767639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,8,power_law_1.01,0.5547520160675049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,8,balanced,0.17495999733606973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,8,balanced,0.21448532740275064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,8,power_law_1.01,0.5729536056518555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,8,power_law_1.01,0.6204480171203614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,8,power_law_1.01,0.7075967788696289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,8,power_law_1.01,0.7868800163269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,8,power_law_1.01,0.3174015998840332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,8,power_law_1.01,1.0807295799255372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,8,power_law_1.01,0.32955520153045653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,8,power_law_1.01,0.3432960033416748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,8,power_law_1.01,1.3118335723876953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,8,power_law_1.01,0.3516607999801636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,8,power_law_1.01,0.3737215995788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,8,power_law_1.01,0.40188798904418943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,8,power_law_1.01,1.9862400054931642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,8,power_law_1.01,0.4296383857727051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,8,power_law_1.01,0.4969791889190674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,8,power_law_1.01,2.3971839904785157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,8,power_law_1.01,3.51673583984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,8,power_law_1.01,0.6124800205230713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,8,balanced,0.2392373283704122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,8,balanced,0.24324800570805868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,8,balanced,0.2469653288523356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,8,balanced,0.27935999631881714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,8,balanced,0.2876426577568054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,8,balanced,0.29783467451731366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,8,balanced,0.30427734057108563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,8,power_law_1.01,4.466239929199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,8,balanced,0.32362133264541626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,8,balanced,0.34190932909647626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,8,balanced,0.35787733395894367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,8,balanced,0.38973331451416016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,8,balanced,0.4513920148213704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,8,power_law_1.01,0.7157695770263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,8,power_law_1.01,1.0038463592529296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,8,power_law_1.01,6.507270050048828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,8,power_law_1.01,1.2198399543762206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,8,power_law_1.01,1.7149120330810548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,8,power_law_1.01,2.2304832458496096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,8,power_law_1.2,0.8578304290771485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,8,power_law_1.01,8.39175033569336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,8,power_law_1.01,0.10120960474014282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,8,power_law_1.01,3.175276756286621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,8,power_law_1.01,0.1667904019355774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,8,power_law_1.01,0.15041279792785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,8,1,power_law_1.2,2.4520896911621093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,8,power_law_1.01,0.14903039932250978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,8,power_law_1.2,0.15230079889297485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,8,power_law_1.2,0.177292799949646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,8,power_law_1.01,4.481119918823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,8,power_law_1.2,0.2582144021987915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,8,1,power_law_1.2,2.3882240295410155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,8,power_law_1.2,0.30957438945770266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,8,power_law_1.2,0.125600004196167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,8,power_law_1.2,0.2294719934463501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,8,power_law_1.2,0.1451647996902466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,8,power_law_1.2,0.4134848117828369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,8,power_law_1.2,0.18908799886703492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,8,power_law_1.2,0.4135807991027832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,8,power_law_1.01,0.17693439722061158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,8,power_law_1.2,0.20247039794921876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,8,power_law_1.2,0.42482562065124513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,8,power_law_1.01,0.20099198818206787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,8,power_law_1.2,0.20125439167022705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,8,power_law_1.2,0.42327041625976564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,8,power_law_1.01,6.252761459350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,8,power_law_1.2,0.3000511884689331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,8,power_law_1.01,0.23011839389801025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,8,power_law_1.2,0.43386240005493165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,8,power_law_1.2,0.3035968065261841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,8,power_law_1.01,0.2357952117919922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,8,power_law_1.2,0.44542717933654785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,8,power_law_1.01,0.23927679061889648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,8,power_law_1.2,0.4585855960845947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,8,power_law_1.01,0.26404480934143065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,8,power_law_1.01,0.27921280860900877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,8,power_law_1.01,0.06032000184059143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,8,power_law_1.01,0.285318398475647
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,8,power_law_1.01,0.07830399870872498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,8,power_law_1.01,0.2941760063171387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,8,power_law_1.01,0.07738239765167236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,8,power_law_1.01,0.3118335962295532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,8,power_law_1.01,0.08611840009689331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,8,power_law_1.01,0.3404416084289551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,8,power_law_1.2,0.3067327976226807
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,8,power_law_1.01,0.08355200290679932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,8,power_law_1.01,0.3637183904647827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,8,power_law_1.01,8.435424041748046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,8,power_law_1.2,0.3144256114959717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,8,power_law_1.01,0.09800959825515747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,8,power_law_1.01,0.41775999069213865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,8,power_law_1.01,0.10494719743728638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,8,power_law_1.01,0.10412160158157349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,8,power_law_1.2,0.10063999891281128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,8,power_law_1.01,0.10520960092544555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,8,power_law_1.2,0.11986559629440308
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,8,power_law_1.01,0.10651520490646363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,8,power_law_1.2,0.1500480055809021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,8,power_law_1.01,0.1072767972946167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,8,power_law_1.2,0.16522239446640014
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,8,power_law_1.01,0.11258879899978638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,8,power_law_1.01,0.1343168020248413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,8,power_law_1.2,0.3163583993911743
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,8,power_law_1.01,0.1277567982673645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,8,power_law_1.2,0.4649983882904053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,8,power_law_1.2,0.3330687999725342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,8,power_law_1.01,0.5231359958648681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,8,power_law_1.2,0.5426239967346191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,8,power_law_1.2,0.33790719509124756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,8,power_law_1.2,0.5308032035827637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,8,power_law_1.2,0.3584768056869507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,8,power_law_1.2,0.5620351791381836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,8,power_law_1.2,0.3931967973709106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,8,power_law_1.2,0.6201663970947265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,8,power_law_1.2,0.15481599569320678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,8,power_law_1.2,0.7218048095703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,8,power_law_1.2,0.18202240467071534
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,8,power_law_1.01,0.12622720003128052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,8,power_law_1.2,0.2197567939758301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,8,power_law_1.2,0.8679167747497558
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,8,power_law_1.01,0.2598400115966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,8,power_law_1.2,0.22147839069366454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,8,power_law_1.01,0.6493055820465088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,8,power_law_1.2,0.23299839496612548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,8,power_law_1.2,1.175507164001465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,8,power_law_1.2,0.2652224063873291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,8,power_law_1.01,0.8636159896850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,8,balanced,0.5257600148518881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,8,power_law_1.2,0.4182079792022705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,8,power_law_1.2,1.3797120094299316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,8,balanced,0.7572480042775472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,8,power_law_1.2,0.43387517929077146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,8,balanced,0.8764426708221436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,8,power_law_1.2,0.5291903972625732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,8,balanced,1.2307573159535725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,8,power_law_1.2,0.6420735836029052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,8,power_law_1.01,0.19911680221557618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,8,balanced,1.5912532806396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,8,power_law_1.2,0.7697408199310303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,8,power_law_1.2,0.2759552001953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,8,power_law_1.2,1.0476415634155274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,8,power_law_1.2,0.28083200454711915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,8,power_law_1.01,1.0600319862365724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,8,power_law_1.2,0.3004096031188965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,8,power_law_1.2,1.3373696327209472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,8,power_law_1.2,2.0333440780639647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,8,power_law_1.2,0.3292223930358887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,8,power_law_1.01,1.4487808227539063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,8,power_law_1.2,0.35576961040496824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,8,power_law_1.01,0.16836479902267457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,8,power_law_1.2,2.0299455642700197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,8,power_law_1.2,0.38392961025238037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,8,power_law_1.01,0.17187199592590333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,8,power_law_1.2,0.42290558815002444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,8,balanced,2.202554702758789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,8,power_law_1.01,0.2000511884689331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,8,power_law_1.2,0.5272064208984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,8,power_law_1.2,2.4921920776367186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,8,power_law_1.01,0.24296960830688477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,8,power_law_1.2,0.6499648094177246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,8,balanced,2.92409610748291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,8,power_law_1.2,0.8582143783569336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,8,power_law_1.2,3.5004352569580077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,8,power_law_1.2,1.1307968139648437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,8,balanced,4.253066698710124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,8,power_law_1.2,5.0309185028076175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,8,power_law_1.2,1.557472038269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,8,power_law_1.01,0.31020801067352294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,8,power_law_1.01,0.39209599494934083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,8,power_law_1.2,4.512339019775391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,8,power_law_1.2,1.9332672119140626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,8,balanced,5.583279927571614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,8,power_law_1.01,0.5357888221740723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,8,1,balanced,0.06159999966621399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,8,1,balanced,0.07619200150171916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,8,1,balanced,0.09700799981753032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,8,1,balanced,0.15659200151761374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,8,1,balanced,0.32598400115966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,8,1,balanced,0.35132265090942383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,8,1,balanced,0.3529493411382039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,8,1,balanced,0.35520001252492267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,8,1,balanced,0.35573331514994305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,8,1,balanced,0.355679988861084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,8,1,balanced,0.3600533405939738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,8,1,balanced,0.3648480176925659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,8,1,balanced,0.41467201709747314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,8,1,balanced,0.42366401354471844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,8,power_law_1.01,1.8613759994506835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,8,1,balanced,0.4317653179168701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,8,power_law_1.2,2.828531265258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,8,1,balanced,0.44077332814534503
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,8,power_law_1.01,0.6961215972900391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,8,power_law_1.01,2.5618688583374025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,8,power_law_1.01,1.009875202178955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,8,power_law_1.01,1.5012031555175782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,8,power_law_1.2,2.290508842468262
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,8,1,power_law_1.2,0.6452288150787353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,8,1,balanced,0.045909335215886436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,8,1,balanced,0.0666240006685257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,8,1,balanced,0.08374933401743571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,8,1,balanced,0.12611732880274454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,8,1,balanced,0.45608532428741455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,8,1,balanced,0.22017600138982138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,8,1,balanced,0.2756426731745402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,8,1,balanced,0.2779253323872884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,8,1,balanced,0.27829333146413165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,8,power_law_1.2,0.05999360084533691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,8,1,balanced,0.27899734179178876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,8,1,balanced,0.28142933050791424
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,8,power_law_1.2,0.06501759886741638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,8,power_law_1.2,0.07333120107650756
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,8,power_law_1.2,0.08554239869117737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,8,power_law_1.2,3.56126708984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,8,power_law_1.2,0.08060799837112427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,8,power_law_1.01,3.5571006774902343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,8,power_law_1.2,0.09576320052146911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,8,1,balanced,0.4893866777420044
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,8,power_law_1.2,0.10110080242156982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,8,1,balanced,0.525984009106954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,8,power_law_1.2,0.10384639501571655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,8,1,balanced,0.7329813639322916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,8,1,balanced,0.2879413366317749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,8,1,balanced,0.7804640134175619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,8,1,balanced,0.29130132993062335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,8,1,balanced,0.30880532662073773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,8,1,balanced,0.3168639938036601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,8,1,balanced,0.32570133606592816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,8,1,balanced,0.3375946680704753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,8,power_law_1.01,5.289676666259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,8,1,balanced,0.35545066992441815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,8,power_law_1.2,7.282335662841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,8,1,balanced,0.3943306605021159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,8,power_law_1.2,5.9403327941894535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,8,1,balanced,0.43798398971557617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,8,1,balanced,0.634170651435852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,8,1,balanced,0.7090773582458496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,8,power_law_1.2,0.10837759971618652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,8,1,balanced,0.9919040203094482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,8,power_law_1.2,0.10962560176849365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,8,power_law_1.2,3.6818241119384765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,8,1,balanced,1.0816853046417236
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,8,power_law_1.2,0.11091840267181396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,8,1,balanced,1.275439977645874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,8,power_law_1.2,0.11324800252914428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,8,1,balanced,1.3625706036885579
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,8,power_law_1.2,0.140665602684021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,8,1,balanced,1.7285386721293132
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,8,power_law_1.2,0.12626559734344484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,8,1,balanced,1.8160853385925293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,8,power_law_1.01,7.61053466796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,8,power_law_1.2,0.1334272027015686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,8,1,balanced,2.294421354929606
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,8,power_law_1.2,0.19329919815063476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,8,1,balanced,2.415818691253662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,8,power_law_1.2,0.19993599653244018
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,8,power_law_1.2,0.17125120162963867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,8,1,balanced,3.311973253885905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,8,power_law_1.2,0.20452480316162108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,8,1,balanced,3.444746653238932
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,8,power_law_1.2,0.2166208028793335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,8,power_law_1.2,0.2515199899673462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,8,power_law_1.2,0.3198591947555542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,8,power_law_1.2,4.8419647216796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,8,1,balanced,4.498154640197754
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,8,power_law_1.2,0.4191232204437256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,8,1,power_law_1.01,0.05969280004501343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,8,1,power_law_1.01,0.07616000175476074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,8,power_law_1.2,7.142668914794922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,8,1,power_law_1.01,0.09539200067520141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,8,1,power_law_1.01,0.13722879886627198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,8,1,power_law_1.01,0.03845759928226471
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,8,1,balanced,0.029578665892283123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,8,1,power_law_1.01,0.18186240196228026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,8,1,power_law_1.01,0.06513280272483826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,8,1,balanced,0.03367999941110611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,8,1,balanced,4.335733413696289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,8,1,power_law_1.01,0.320467209815979
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,8,power_law_1.2,0.6020671844482421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,8,1,power_law_1.01,0.3309056043624878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,8,power_law_1.2,7.350003051757812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,8,1,power_law_1.01,0.3385535955429077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,8,1,power_law_1.01,0.3359935998916626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,8,1,power_law_1.01,0.3397056102752686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,8,1,power_law_1.01,0.3460416078567505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,8,1,power_law_1.01,0.3800447940826416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,8,1,balanced,0.04523199796676636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,8,1,power_law_1.01,0.08393599987030029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,8,1,power_law_1.01,0.40145277976989746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,8,1,power_law_1.01,0.1176576018333435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,8,1,power_law_1.01,0.4174464225769043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,8,1,power_law_1.01,0.14709119796752929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,8,1,power_law_1.01,0.4398655891418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,8,power_law_1.2,0.7565311908721923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,8,1,power_law_1.01,0.219980788230896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,8,1,power_law_1.01,0.43628802299499514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,8,1,power_law_1.01,0.22548480033874513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,8,1,power_law_1.01,0.5182208061218262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,8,power_law_1.2,10.670829010009765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,8,1,balanced,0.06512000163396199
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,8,1,balanced,0.09557333588600159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,8,1,power_law_1.2,0.058195197582244874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,8,1,power_law_1.2,0.07271680235862732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,8,1,power_law_1.2,0.09379839897155762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,8,1,power_law_1.2,0.12876800298690796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,8,1,power_law_1.2,0.17000319957733154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,8,1,power_law_1.01,0.23592960834503174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,8,1,power_law_1.2,0.3058111906051636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,8,1,power_law_1.01,0.2486016035079956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,8,1,power_law_1.01,0.5546688079833985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,8,1,power_law_1.01,0.2616447925567627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,8,1,power_law_1.01,0.027513599395751952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,8,1,balanced,0.1127893328666687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,8,1,power_law_1.01,0.5959871768951416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,8,1,power_law_1.01,0.2866111993789673
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,8,1,power_law_1.01,0.029683199524879456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,8,1,power_law_1.01,0.3108223915100098
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,8,1,power_law_1.01,0.044947201013565065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,8,1,power_law_1.01,0.30098559856414797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,8,1,power_law_1.01,0.05619199872016907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,8,1,power_law_1.01,0.317792010307312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,8,1,power_law_1.01,0.07146880030632019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,8,1,power_law_1.01,0.3297728061676025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,8,1,power_law_1.01,0.09757440090179444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,8,1,power_law_1.01,0.10278400182723998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,8,1,power_law_1.01,0.10967040061950684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,8,1,balanced,0.11691199739774068
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,8,1,power_law_1.01,0.11560959815979004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,8,1,balanced,0.12071466445922852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,8,power_law_1.2,9.755417633056641
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,8,1,balanced,0.12302933136622111
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,8,1,power_law_1.01,0.1193727970123291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,8,1,balanced,0.12541866302490234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,8,1,power_law_1.01,0.7007872104644776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,8,1,balanced,0.13053866227467856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,8,1,power_law_1.2,0.03850240111351013
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,8,1,power_law_1.01,0.1259392023086548
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,8,1,balanced,0.13568533460299173
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,8,1,balanced,0.15012799700101218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,8,1,power_law_1.2,0.06702719926834107
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,8,1,power_law_1.01,0.1327232003211975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,8,1,power_law_1.2,0.08330879807472229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,8,1,power_law_1.01,0.14065279960632324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,8,power_law_1.2,1.0677056312561035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,8,1,power_law_1.2,0.10709760189056397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,8,1,power_law_1.01,0.15830399990081787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,8,1,power_law_1.2,0.32356479167938235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,8,1,power_law_1.2,0.1410688042640686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,8,1,power_law_1.2,0.32791039943695066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,8,1,power_law_1.2,0.21880319118499755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,8,1,power_law_1.2,0.8688703536987304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,8,1,power_law_1.2,0.3320255994796753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,8,1,power_law_1.2,0.2292543888092041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,8,1,power_law_1.2,0.3387264013290405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,8,1,power_law_1.2,0.243174409866333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,8,1,power_law_1.2,0.3603327989578247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,8,1,power_law_1.2,0.2562560081481934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,8,1,power_law_1.01,0.8384256362915039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,8,1,power_law_1.2,0.3882944107055664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,8,1,power_law_1.2,0.40670080184936525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,8,1,power_law_1.01,1.0933631896972655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,8,1,power_law_1.01,0.34673919677734377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,8,1,power_law_1.2,0.41890559196472166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,8,1,power_law_1.01,0.17446399927139283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,8,1,power_law_1.01,0.38487679958343507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,8,1,power_law_1.2,0.42942080497741697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,8,1,power_law_1.01,1.351078414916992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,8,1,power_law_1.01,0.4526656150817871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,8,1,power_law_1.2,0.44581117630004885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,8,1,power_law_1.01,0.5083903789520263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,8,1,power_law_1.2,0.5195903778076172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,8,1,balanced,0.16180800398190817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,8,1,power_law_1.01,0.6269887924194336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,8,1,power_law_1.2,0.26883840560913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,8,1,power_law_1.2,0.560588788986206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,8,1,power_law_1.2,0.2954623937606812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,8,1,power_law_1.2,0.6027775764465332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,8,1,power_law_1.2,0.31468160152435304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,8,1,power_law_1.2,0.3086143970489502
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,8,1,power_law_1.01,0.15459840297698973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,8,1,power_law_1.2,0.32051839828491213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,8,1,power_law_1.01,0.17552640438079833
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,8,1,power_law_1.2,0.028428798913955687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,8,1,power_law_1.01,0.18513280153274536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,8,1,power_law_1.2,0.03175680041313171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,8,1,power_law_1.01,1.8949119567871093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,8,1,balanced,0.16991466283798218
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,8,1,power_law_1.2,0.045440000295639035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,8,1,power_law_1.01,0.7575295925140381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,8,1,power_law_1.2,0.7181632041931152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,8,1,power_law_1.01,1.0106304168701172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,8,1,power_law_1.2,0.8454848289489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,8,1,power_law_1.2,0.33729920387268064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,8,1,power_law_1.01,1.2748607635498046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,8,1,power_law_1.2,0.3501440048217773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,8,1,power_law_1.01,0.21175038814544678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,8,1,power_law_1.2,1.1020799636840821
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,8,1,balanced,0.1618613302707672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,8,1,power_law_1.01,0.25253119468688967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,8,1,balanced,0.17706133921941122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,8,1,balanced,0.1908479928970337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,8,1,power_law_1.01,0.3020096063613892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,8,1,balanced,0.20946667591730753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,8,1,balanced,0.2988693316777547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,8,1,power_law_1.01,2.414259147644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,8,1,power_law_1.01,0.36059520244598386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,8,1,balanced,0.2877013285954793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,8,1,power_law_1.2,1.3778495788574219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,8,1,balanced,0.6404853264490763
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,8,1,power_law_1.01,0.43767681121826174
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,8,1,balanced,0.518671989440918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,8,1,balanced,0.7423893610636393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,8,1,power_law_1.2,1.907481575012207
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,8,1,balanced,0.9703359603881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,8,1,power_law_1.01,3.4901695251464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,8,1,power_law_1.2,0.05426560044288635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,8,1,balanced,1.0409226417541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,8,1,power_law_1.2,0.07009279727935791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,8,1,power_law_1.2,2.4358911514282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,8,1,balanced,1.4573173522949219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,8,balanced,0.07740800082683563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,8,balanced,0.09201600154240926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,8,balanced,0.1093280017375946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,8,balanced,0.1667893330256144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,8,balanced,0.30450133482615155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,8,1,power_law_1.01,0.5985087871551513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,8,1,power_law_1.01,4.623257446289062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,8,1,power_law_1.2,0.38567678928375243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,8,balanced,0.05384533107280731
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,8,power_law_1.2,1.7260160446166992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,8,balanced,0.08029333253701527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,8,1,power_law_1.01,0.75764479637146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,8,1,power_law_1.01,1.7984064102172852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,8,1,power_law_1.2,0.45705599784851075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,8,1,power_law_1.2,0.5082880020141601
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,8,1,power_law_1.01,1.0672896385192872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,8,1,power_law_1.2,0.6365568161010742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,8,balanced,0.07092800239721934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,8,balanced,0.08429867029190063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,8,balanced,0.10217066605885823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,8,balanced,0.1234933336575826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,8,1,power_law_1.01,1.4451328277587892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,8,balanced,0.20796799659729004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,8,balanced,0.5297813415527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,8,balanced,0.3986826737721761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,8,balanced,0.7618559996287028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,8,balanced,0.5845866600672404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,8,balanced,0.09701333443323772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,8,balanced,0.9495626290639242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,8,balanced,0.13218667109807333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,8,balanced,0.2060319979985555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,8,balanced,0.953167994817098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,8,balanced,0.35124798615773517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,8,balanced,0.5161759853363037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,8,1,power_law_1.2,0.0971455991268158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,8,balanced,0.9855360190073649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,8,balanced,0.7151626745859782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,8,1,power_law_1.2,0.10440959930419921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,8,1,power_law_1.01,2.3187904357910156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,8,balanced,0.7215466499328613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,8,1,power_law_1.2,0.11296000480651855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,8,1,power_law_1.2,0.7656383991241456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,8,balanced,0.7261599699656168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,8,1,power_law_1.2,0.11521919965744018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,8,balanced,0.738368034362793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,8,1,power_law_1.2,0.12238080501556396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,8,balanced,0.7469173272450765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,8,1,power_law_1.2,0.12684160470962524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,8,balanced,0.5878133376439413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,8,balanced,0.7589866320292155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,8,balanced,0.5926773150761923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,8,1,power_law_1.01,3.373126220703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,8,balanced,0.7774453163146973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,8,balanced,0.596837321917216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,8,balanced,0.7997013727823893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,8,balanced,0.6058186690012614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,8,balanced,0.817909320195516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,8,balanced,0.9696479638417562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,8,balanced,0.8544480005900065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,8,balanced,0.9762453238169352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,8,balanced,0.9239786465962728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,8,balanced,0.9734559853871664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,8,balanced,1.000816027323405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,8,balanced,1.005130688349406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,8,balanced,1.145967960357666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,8,balanced,1.0110507011413574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,8,1,power_law_1.2,0.136627197265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,8,balanced,1.3004319667816162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,8,balanced,0.6160266796747843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,8,balanced,1.604533354441325
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,8,balanced,0.029466666281223297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,8,balanced,0.6264319817225138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,8,balanced,0.03160000095764796
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,8,balanced,0.039408000806967415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,8,balanced,0.0635040005048116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,8,balanced,0.09106133381525676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,8,1,power_law_1.2,3.5108543395996095
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,8,balanced,0.14666666587193808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,8,balanced,2.187994639078776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,8,balanced,0.20329066117604574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,8,1,power_law_1.2,1.014022445678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,8,balanced,0.20780267318089804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,8,1,power_law_1.01,4.417279815673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,8,balanced,0.20939733584721884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,8,balanced,0.21166932582855225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,8,balanced,0.207696000734965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,8,power_law_1.01,0.15168639421463012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,8,1,power_law_1.2,1.2852160453796386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,8,balanced,2.7945706049601235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,8,balanced,1.0506133238474529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,8,power_law_1.01,0.3019007921218872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,8,balanced,1.0834186871846516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,8,1,power_law_1.2,1.8043455123901366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,8,power_law_1.01,0.52259202003479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,8,balanced,0.640992005666097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,8,balanced,3.685482660929362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,8,balanced,0.6621866623560587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,8,power_law_1.01,0.4718143939971924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,8,balanced,0.6782453060150146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,8,balanced,0.7118026415506998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,8,power_law_1.01,0.5246272087097168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,8,balanced,0.7736159960428873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,8,balanced,0.2130933403968811
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,8,1,power_law_1.2,0.14450559616088868
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,8,balanced,0.2182240088780721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,8,balanced,0.8358240127563477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,8,balanced,0.21797333161036173
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,8,1,power_law_1.2,0.1624575972557068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,8,balanced,5.181359926859538
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,8,balanced,0.22295467058817545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,8,balanced,0.9586346944173177
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,8,1,power_law_1.2,0.1789888024330139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,8,balanced,1.1356106599171956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,8,balanced,1.0813653469085693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,8,1,power_law_1.2,2.330009651184082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,8,balanced,1.190399964650472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,8,balanced,1.3445332845052083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,8,balanced,1.3202293713887532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,8,balanced,1.798495928446452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,8,balanced,1.458565394083659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,8,balanced,7.091226577758789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,8,power_law_1.01,0.6895936012268067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,8,balanced,1.7491733233133953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,8,balanced,0.228112002213796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,8,balanced,2.3099679946899414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,8,power_law_1.01,0.11584639549255371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,8,power_law_1.01,0.6748415946960449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,8,power_law_1.01,0.19804160594940184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,8,balanced,2.3111093839009604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,8,balanced,3.0467093785603843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,8,power_law_1.01,0.8889856338500977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,8,power_law_1.01,0.32707839012145995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,8,balanced,2.8589760462443032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,8,power_law_1.01,0.8653887748718262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,8,balanced,4.2975413004557295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,8,power_law_1.01,0.8524479866027832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,8,balanced,0.2909653385480245
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,8,1,power_law_1.2,0.15834239721298218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,8,balanced,3.8607199986775718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,8,1,power_law_1.2,0.17895040512084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,8,power_law_1.01,0.8667648315429688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,8,1,power_law_1.2,0.18965760469436646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,8,1,power_law_1.2,0.2171328067779541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,8,power_law_1.01,0.8510272026062011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,8,power_law_1.01,0.30950400829315183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,8,1,power_law_1.2,0.2579967975616455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,8,balanced,5.80619748433431
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,8,power_law_1.01,0.3673408031463623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,8,1,power_law_1.2,0.31432960033416746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,8,balanced,5.824538548787435
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,8,1,power_law_1.2,0.3649343967437744
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,8,1,power_law_1.2,0.4460031986236572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,8,power_law_1.01,0.107315194606781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,8,1,power_law_1.2,0.6095168113708496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,8,power_law_1.01,0.18259840011596679
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,8,1,power_law_1.2,0.7639167785644532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,8,power_law_1.01,0.3267008066177368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,8,power_law_1.01,0.8990336418151855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,8,balanced,8.669616063435873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,8,power_law_1.01,0.4108096122741699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,8,power_law_1.01,0.9105600357055664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,8,power_law_1.01,0.43837437629699705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,8,power_law_1.01,0.9514176368713378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,8,power_law_1.01,0.685964822769165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,8,power_law_1.01,1.0077631950378418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,8,power_law_1.01,0.6667456150054931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,8,power_law_1.01,0.3023616075515747
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,8,balanced,0.2637066642443339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,8,power_law_1.01,0.6541120052337647
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,8,balanced,0.5856800079345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,8,power_law_1.01,0.3997247934341431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,8,balanced,0.3614559968312581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,8,power_law_1.01,0.6876736164093018
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,8,balanced,0.3203360040982564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,8,balanced,0.3548479874928792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,8,power_law_1.01,0.5039040088653565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,8,power_law_1.01,0.6753920078277588
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,8,balanced,0.3956640164057414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,8,balanced,0.4840799967447917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,8,power_law_1.01,0.5172800064086914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,8,power_law_1.01,0.7138815879821777
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,8,balanced,0.667733351389567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,8,power_law_1.01,1.0094528198242188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,8,1,power_law_1.2,1.0713664054870606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,8,power_law_1.01,0.5529215812683106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,8,power_law_1.01,0.5393856048583985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,8,power_law_1.01,1.1394495964050293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,8,power_law_1.01,0.5594880104064941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,8,power_law_1.01,1.2296128273010254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,8,1,power_law_1.2,3.3922176361083984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,8,power_law_1.01,0.5220352172851562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,8,power_law_1.01,1.4470080375671386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,8,power_law_1.01,0.7445184230804444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,8,power_law_1.01,0.7591872215270996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,8,power_law_1.01,1.6655359268188477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,8,power_law_1.01,0.7777664184570312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,8,power_law_1.01,0.8327808380126953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,8,power_law_1.01,2.0730432510375976
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,8,power_law_1.01,0.056908798217773435
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,8,balanced,0.8276693026224772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,8,power_law_1.01,0.9268287658691406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,8,power_law_1.01,0.0840511977672577
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,8,balanced,1.1677813529968262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,8,power_law_1.01,2.4930240631103517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,8,power_law_1.01,1.037183952331543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,8,power_law_1.01,0.13746559619903564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,8,power_law_1.2,0.15208319425582886
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,8,power_law_1.01,0.1316864013671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,8,power_law_1.2,0.3027071952819824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,8,power_law_1.01,1.2868800163269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,8,power_law_1.01,0.1455296039581299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,8,power_law_1.2,0.3924031972885132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,8,power_law_1.01,0.5726912021636963
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,8,power_law_1.01,0.17958400249481202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,8,power_law_1.01,1.5324352264404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,8,power_law_1.2,0.4712575912475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,8,power_law_1.01,0.15763200521469117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,8,power_law_1.2,0.5552832126617432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,8,power_law_1.01,0.19901440143585206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,8,1,power_law_1.2,4.440927886962891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,8,power_law_1.01,1.9883136749267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,8,power_law_1.01,0.17676160335540772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,8,power_law_1.2,0.646233606338501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,8,power_law_1.2,0.11687040328979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,8,power_law_1.01,0.19748480319976808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,8,power_law_1.2,0.1987712025642395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,8,power_law_1.2,0.6204095840454101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,8,power_law_1.01,2.4266944885253907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,8,power_law_1.01,0.18273919820785522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,8,power_law_1.2,0.26232318878173827
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,8,power_law_1.01,0.18700159788131715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,8,power_law_1.2,0.9018943786621094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,8,power_law_1.2,0.30230400562286375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,8,power_law_1.01,0.18849920034408568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,8,power_law_1.2,0.8591232299804688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,8,power_law_1.2,0.40265598297119143
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,8,power_law_1.01,0.19635839462280275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,8,power_law_1.01,3.4063167572021484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,8,power_law_1.01,0.19916160106658937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,8,power_law_1.2,0.4511551856994629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,8,power_law_1.2,0.8177215576171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,8,power_law_1.01,0.21622400283813475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,8,power_law_1.01,3.484819030761719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,8,power_law_1.2,0.4769279956817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,8,1,power_law_1.2,4.5980480194091795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,8,power_law_1.01,0.2362112045288086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,8,power_law_1.2,0.8958656311035156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,8,power_law_1.01,0.5533120155334472
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,8,power_law_1.01,0.24296319484710693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,8,power_law_1.2,0.6581503868103027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,8,power_law_1.01,4.211468887329102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,8,power_law_1.2,0.8649663925170898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,8,power_law_1.01,0.5510399818420411
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,8,power_law_1.01,0.37392001152038573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,8,power_law_1.2,0.6594880104064942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,8,power_law_1.01,0.601580810546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,8,power_law_1.01,0.31548800468444826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,8,power_law_1.01,4.523532867431641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,8,power_law_1.2,0.6517951965332032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,8,power_law_1.2,0.10795520544052124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,8,power_law_1.01,0.6192831993103027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,8,power_law_1.2,0.18249599933624266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,8,power_law_1.2,0.6872704029083252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,8,power_law_1.01,0.677888011932373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,8,power_law_1.2,0.24840319156646729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,8,power_law_1.2,0.6730751991271973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,8,power_law_1.01,0.7723775863647461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,8,power_law_1.2,0.32498559951782224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,8,power_law_1.01,6.2537792205810545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,8,power_law_1.2,0.9092608451843261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,8,power_law_1.2,0.6753536224365234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,8,power_law_1.01,0.8439488410949707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,8,power_law_1.01,6.329030227661133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,8,power_law_1.01,0.3158143997192383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,8,power_law_1.2,0.9447808265686035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,8,power_law_1.2,0.7334015846252442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,8,power_law_1.01,1.0327679634094238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,8,power_law_1.01,0.38309760093688966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,8,power_law_1.2,0.7462592124938965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,8,power_law_1.01,1.1936511993408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,8,power_law_1.2,0.788153600692749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,8,power_law_1.2,0.37087359428405764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,8,power_law_1.01,1.5743743896484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,8,power_law_1.01,7.880646514892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,8,power_law_1.2,0.4844992160797119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,8,power_law_1.01,1.9264768600463866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,8,power_law_1.2,0.9791616439819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,8,power_law_1.2,0.5269696235656738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,8,power_law_1.01,8.438553619384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,8,power_law_1.01,0.45119361877441405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,8,1,balanced,0.06673066814740498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,8,power_law_1.2,1.0008000373840331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,8,1,balanced,0.07878399888674419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,8,power_law_1.2,0.5864384174346924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,8,1,balanced,0.09916266798973083
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,8,power_law_1.01,0.5880832195281982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,8,power_law_1.2,0.8428031921386718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,8,power_law_1.01,2.6622400283813477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,8,power_law_1.2,1.0453887939453126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,8,power_law_1.2,0.5608960151672363
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,8,power_law_1.01,0.7377024173736573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,8,power_law_1.2,0.9348544120788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,8,power_law_1.2,0.4836863994598389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,8,power_law_1.2,1.1323136329650878
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,8,power_law_1.01,1.0514944076538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,8,power_law_1.2,1.0843199729919433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,8,power_law_1.2,1.2431167602539062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,8,1,balanced,0.14596266547838846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,8,power_law_1.2,1.3011072158813477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,8,1,balanced,0.3157973289489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,8,1,balanced,0.5566346645355225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,8,1,balanced,0.7826826572418213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,8,power_law_1.2,1.5599040031433105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,8,1,balanced,0.7760480244954427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,8,1,balanced,0.7806186676025391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,8,power_law_1.2,0.525708818435669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,8,power_law_1.2,2.0173248291015624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,8,power_law_1.2,0.5559936046600342
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,8,power_law_1.01,1.7455232620239258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,8,power_law_1.2,1.4663871765136718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,8,power_law_1.2,0.5507328033447265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,8,1,balanced,0.04192000130812327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,8,1,balanced,0.07022933165232341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,8,power_law_1.2,2.4724544525146483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,8,1,balanced,0.0904960036277771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,8,1,balanced,0.1222826639811198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,8,1,balanced,0.22671467065811157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,8,1,balanced,0.38982399304707843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,8,power_law_1.2,0.05596799850463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,8,power_law_1.2,1.7154239654541015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,8,1,balanced,0.5762773354848226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,8,1,balanced,0.7820106347401937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,8,power_law_1.2,0.08353279829025269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,8,1,balanced,0.578384002049764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,8,power_law_1.01,3.4350208282470702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,8,1,balanced,0.5785013437271118
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,8,power_law_1.2,0.09867519736289979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,8,1,balanced,0.581551988919576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,8,power_law_1.2,2.2287296295166015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,8,1,balanced,0.5843466520309448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,8,power_law_1.2,0.12363519668579101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,8,1,balanced,0.5903840065002441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,8,1,balanced,0.5960479974746704
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,8,power_law_1.2,0.150271999835968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,8,1,balanced,0.6030986706415812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,8,power_law_1.2,0.543833589553833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,8,1,balanced,0.6120746533075968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,8,power_law_1.2,2.7527360916137695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,8,power_law_1.2,0.19050240516662598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,8,1,balanced,0.6218719879786173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,8,power_law_1.2,0.6081664085388183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,8,power_law_1.01,5.044409561157226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,8,1,balanced,0.6852693557739258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,8,power_law_1.2,0.1996351957321167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,8,power_law_1.2,0.6321280002593994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,8,power_law_1.2,3.491571044921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,8,power_law_1.2,0.18862719535827638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,8,power_law_1.2,0.6780543804168702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,8,power_law_1.2,3.7969921112060545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,8,power_law_1.2,0.20632319450378417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,8,power_law_1.2,0.7573247909545898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,8,power_law_1.2,0.17357439994812013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,8,power_law_1.2,0.8704128265380859
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,8,power_law_1.2,0.1707584023475647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,8,power_law_1.01,6.51435546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,8,1,balanced,0.7234079837799072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,8,power_law_1.2,1.0675071716308593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,8,power_law_1.2,0.18284800052642822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,8,1,balanced,0.7652906576792399
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,8,power_law_1.2,0.1918463945388794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,8,power_law_1.2,1.2463168144226073
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,8,power_law_1.2,0.18896000385284423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,8,power_law_1.2,0.20129919052124023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,8,power_law_1.2,1.6158592224121093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,8,1,power_law_1.2,1.4844672203063964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,8,power_law_1.2,4.387795257568359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,8,1,balanced,0.050240000089009605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,8,1,balanced,0.060085331400235496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,8,power_law_1.2,2.117523193359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,8,power_law_1.2,4.631935882568359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,8,power_law_1.2,2.93756160736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,8,1,balanced,0.7865066528320312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,8,1,balanced,0.7927253246307373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,8,power_law_1.2,6.733567810058593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,8,1,balanced,0.7917439937591553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,8,1,balanced,0.07411733269691467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,8,1,balanced,0.10315733154614766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,8,1,balanced,0.7923519611358643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,8,1,balanced,0.172650674978892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,8,1,balanced,0.3163999915122986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,8,1,balanced,0.8038826783498129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,8,power_law_1.2,6.516287994384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,8,1,balanced,0.4476586580276489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,8,power_law_1.2,3.917388916015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,8,1,balanced,0.808847983678182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,8,1,balanced,0.44526398181915283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,8,1,balanced,0.44897600015004474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,8,1,balanced,0.9142133394877116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,8,1,balanced,0.4503200054168701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,8,power_law_1.2,0.20894079208374022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,8,1,balanced,0.45955200990041095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,8,1,balanced,0.46532801787058514
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,8,power_law_1.2,0.2332223892211914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,8,1,balanced,0.4718720118204753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,8,power_law_1.2,0.2477247953414917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,8,1,balanced,0.46926931540171307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,8,1,balanced,0.8447413444519043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,8,1,balanced,0.477349321047465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,8,power_law_1.2,0.3816447973251343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,8,1,balanced,0.48599998156229657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,8,power_law_1.2,8.237235260009765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,8,power_law_1.2,5.416025543212891
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,8,power_law_1.2,0.3173504114151001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,8,1,power_law_1.01,0.062035202980041504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,8,power_law_1.2,0.37457919120788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,8,1,power_law_1.01,0.07708160281181335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,8,1,balanced,0.9732267061869303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,8,power_law_1.2,9.098809814453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,8,1,balanced,0.9694293340047201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,8,1,power_law_1.01,0.09835519790649414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,8,1,balanced,1.0386772950490315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,8,1,power_law_1.01,0.039628800749778745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,8,1,power_law_1.01,0.14431999921798705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,8,1,balanced,1.1163360277811687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,8,1,balanced,0.9283680121103922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,8,1,power_law_1.01,0.28377599716186525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,8,1,balanced,0.503221352895101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,8,1,balanced,1.293178637822469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,8,1,balanced,0.5307519833246866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,8,1,power_law_1.01,0.38429439067840576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,8,1,balanced,0.6635040044784546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,8,1,balanced,0.030581332743167877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,8,1,balanced,1.7645227114359539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,8,1,balanced,0.7122986316680908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,8,1,power_law_1.01,0.4245312213897705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,8,power_law_1.2,0.4242432117462158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,8,1,balanced,0.9102559884389242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,8,1,balanced,2.0454079310099282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,8,1,power_law_1.01,0.5513535976409912
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,8,power_law_1.2,0.4955904006958008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,8,1,balanced,1.0582773685455322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,8,1,power_law_1.01,0.05811200141906738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,8,1,power_law_1.01,0.5671743869781494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,8,1,balanced,2.7046934763590493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,8,power_law_1.2,0.648038387298584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,8,1,balanced,1.5547199249267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,8,1,power_law_1.01,0.0868224024772644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,8,1,balanced,1.1078879833221436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,8,1,power_law_1.01,0.5891520023345947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,8,1,power_law_1.01,0.11921919584274292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,8,power_law_1.2,7.0487617492675785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,8,power_law_1.2,0.7967103958129883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,8,1,balanced,1.8092373212178547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,8,1,balanced,1.5472745895385742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,8,1,power_law_1.01,0.6014912128448486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,8,1,balanced,3.6655146280924478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,8,1,balanced,1.898789405822754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,8,1,balanced,2.426426728566488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,8,1,balanced,2.4995840390523276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,8,1,balanced,3.3503573735555015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,8,1,balanced,5.558090845743815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,8,1,balanced,0.032101333141326904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,8,1,balanced,0.04182933270931244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,8,1,balanced,0.0639573335647583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,8,1,power_law_1.01,0.2103679895401001
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,8,1,balanced,0.09521599610646565
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,8,1,balanced,0.15520000457763672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,8,1,balanced,0.21938133239746094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,8,1,power_law_1.01,0.2824959993362427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,8,1,balanced,0.22076799472173056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,8,1,power_law_1.01,0.6166912078857422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,8,1,balanced,0.22212799390157065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,8,1,balanced,4.62282657623291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,8,1,balanced,0.22563733657201132
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,8,1,balanced,0.2302346626917521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,8,1,power_law_1.01,0.6337408065795899
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,8,1,balanced,0.23643734057744345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,8,1,power_law_1.2,0.0639680027961731
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,8,1,balanced,0.24335465828577676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,8,1,balanced,0.2516480088233948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,8,1,power_law_1.01,0.6831808090209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,8,1,power_law_1.01,0.727942419052124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,8,1,power_law_1.01,0.04928640127182007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,8,1,power_law_1.01,0.7628736019134521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,8,1,power_law_1.01,0.060313600301742556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,8,1,power_law_1.01,0.33225600719451903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,8,1,power_law_1.01,0.07166720032691956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,8,1,power_law_1.01,0.8479104042053223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,8,1,power_law_1.01,0.10147839784622192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,8,1,power_law_1.01,0.40082559585571287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,8,1,power_law_1.2,0.07702400088310242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,8,1,power_law_1.01,0.9545856475830078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,8,1,power_law_1.01,0.16202239990234374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,8,1,power_law_1.2,0.09781119823455811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,8,1,power_law_1.01,0.4244800090789795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,8,1,power_law_1.01,0.9919551849365235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,8,1,power_law_1.2,0.1440832018852234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,8,1,balanced,3.4900639851888022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,8,1,power_law_1.01,0.45167999267578124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,8,power_law_1.2,1.1193792343139648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,8,1,power_law_1.2,0.22728960514068602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,8,1,power_law_1.01,1.100153636932373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,8,1,power_law_1.2,0.3493119955062866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,8,1,power_law_1.01,1.212883186340332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,8,1,balanced,4.682661374409993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,8,1,power_law_1.2,0.40945281982421877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,8,1,power_law_1.01,0.02901119887828827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,8,1,power_law_1.2,0.5253376007080078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,8,1,power_law_1.01,1.4659584045410157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,8,1,power_law_1.01,0.2405440092086792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,8,1,power_law_1.01,0.030847999453544616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,8,1,power_law_1.2,0.5715968132019043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,8,1,power_law_1.01,0.0401856005191803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,8,1,power_law_1.01,1.729804801940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,8,1,power_law_1.2,0.5728511810302734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,8,1,balanced,0.26174400250116986
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,8,power_law_1.2,2.0378816604614256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,8,1,balanced,0.27158933877944946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,8,1,power_law_1.2,0.5886784076690674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,8,1,balanced,0.3234293262163798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,8,1,power_law_1.2,0.04068480134010315
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,8,1,balanced,0.3655253251393636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,8,1,balanced,0.3547840118408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,8,1,power_law_1.2,0.06423680186271667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,8,1,balanced,0.4092106819152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,8,1,balanced,0.43187201023101807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,8,1,power_law_1.01,0.26549758911132815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,8,1,power_law_1.01,0.45525121688842773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,8,1,balanced,0.5175413290659586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,8,1,power_law_1.01,0.3146944046020508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,8,1,power_law_1.01,0.06101760268211365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,8,1,power_law_1.01,0.479801607131958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,8,1,power_law_1.01,0.32193920612335203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,8,1,power_law_1.01,0.08567039966583252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,8,1,power_law_1.01,0.49915518760681155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,8,1,power_law_1.01,2.2696319580078126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,8,1,power_law_1.2,0.6134719848632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,8,1,power_law_1.01,0.330182409286499
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,8,1,power_law_1.01,0.12008320093154908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,8,1,power_law_1.2,0.08836479783058167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,8,1,power_law_1.01,0.5400896072387695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,8,1,power_law_1.2,0.6294015884399414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,8,1,power_law_1.01,0.1486207962036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,8,1,power_law_1.2,0.12061439752578736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,8,1,balanced,0.6410719950993856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,8,1,power_law_1.01,0.5924287796020508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,8,1,power_law_1.2,0.6951488018035888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,8,1,balanced,0.7367839813232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,8,1,power_law_1.01,0.177510404586792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,8,1,power_law_1.2,0.20091519355773926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,8,1,power_law_1.01,0.6392704010009765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,8,1,balanced,1.1686293284098308
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,8,1,power_law_1.01,0.18131200075149537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,8,1,power_law_1.2,0.2610431909561157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,8,1,power_law_1.01,0.6546048164367676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,8,1,balanced,1.0868746439615886
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,8,1,power_law_1.01,0.18533120155334473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,8,1,power_law_1.2,0.30293118953704834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,8,1,power_law_1.01,0.7169151782989502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,8,1,power_law_1.2,0.04844799935817719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,8,1,power_law_1.01,0.3519743919372559
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,8,1,balanced,1.5821545918782551
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,8,1,power_law_1.01,0.1966271996498108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,8,1,power_law_1.2,0.39444479942321775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,8,1,power_law_1.2,0.05894399881362915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,8,1,power_law_1.01,0.3823168039321899
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,8,1,power_law_1.01,0.20726399421691893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,8,1,power_law_1.2,0.4101439952850342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,8,1,power_law_1.2,0.07132160067558288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,8,1,power_law_1.01,2.7822784423828124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,8,1,power_law_1.01,0.39784319400787355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,8,1,power_law_1.01,0.21558399200439454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,8,1,power_law_1.2,0.7309887886047364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,8,1,power_law_1.2,0.43333120346069337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,8,1,power_law_1.01,0.4136960029602051
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,8,1,power_law_1.01,0.2322688102722168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,8,1,power_law_1.2,0.7906623840332031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,8,1,power_law_1.2,0.4580992221832275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,8,1,power_law_1.01,0.2468480110168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,8,1,power_law_1.2,0.4867392063140869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,8,1,power_law_1.2,0.8499711990356446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,8,1,power_law_1.01,3.8563968658447267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,8,1,power_law_1.01,0.7860032081604004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,8,1,power_law_1.2,0.503001594543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,8,1,power_law_1.2,0.967404842376709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,8,1,power_law_1.01,0.9127424240112305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,8,1,power_law_1.2,0.5375167846679687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,8,1,power_law_1.2,0.1008255958557129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,8,1,power_law_1.2,1.0131072044372558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,8,1,power_law_1.2,0.5988224029541016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,8,1,power_law_1.2,0.1472767949104309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,8,1,power_law_1.01,1.0438847541809082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,8,1,power_law_1.01,0.4380159854888916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,8,1,power_law_1.2,0.20675840377807617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,8,1,power_law_1.01,5.067756652832031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,8,1,power_law_1.01,0.2638848066329956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,8,1,power_law_1.01,1.285267162322998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,8,1,power_law_1.01,0.4554111957550049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,8,balanced,0.03370666752258936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,8,balanced,0.04002666721741358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,8,balanced,0.06229333579540253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,8,balanced,0.07635733485221863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,8,1,power_law_1.01,0.2980031967163086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,8,balanced,0.11100799838701884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,8,1,power_law_1.01,0.5210624217987061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,8,balanced,0.19317867358525595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,8,balanced,0.2701813379923503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,8,1,power_law_1.01,1.5498880386352538
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,8,1,power_law_1.01,0.3756095886230469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,8,balanced,0.27020267645517987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,8,balanced,0.27320533990859985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,8,balanced,0.2709439992904663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,8,1,power_law_1.01,0.33556480407714845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,8,balanced,0.27729066212972003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,8,balanced,0.2834399938583374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,8,balanced,0.287663996219635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,8,1,power_law_1.2,1.1089344024658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,8,1,power_law_1.01,0.42741122245788576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,8,balanced,0.2999946673711141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,8,balanced,0.3119306763013204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,8,balanced,0.3200160066286723
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,8,1,power_law_1.01,0.418720006942749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,8,balanced,0.34228265285491943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,8,1,power_law_1.2,0.02905600070953369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,8,1,power_law_1.2,1.25797119140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,8,1,power_law_1.2,0.2534143924713135
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,8,1,power_law_1.2,0.03076480031013489
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,8,1,power_law_1.01,0.547814416885376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,8,1,power_law_1.2,0.3027328014373779
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,8,1,power_law_1.2,0.04068480134010315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,8,1,power_law_1.2,1.493619155883789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,8,1,power_law_1.01,0.6104767799377442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,8,1,power_law_1.2,0.31658239364624025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,8,1,power_law_1.2,0.06159359812736511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,8,1,power_law_1.01,0.7594240188598633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,8,1,power_law_1.01,2.065299224853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,8,1,power_law_1.2,0.32872960567474363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,8,1,power_law_1.2,1.7646144866943358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,8,1,power_law_1.2,0.6462975978851319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,8,1,power_law_1.01,0.9384960174560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,8,1,power_law_1.2,0.34235520362854005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,8,1,power_law_1.2,0.661414384841919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,8,1,power_law_1.2,0.3642944097518921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,8,1,power_law_1.01,2.6034303665161134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,8,1,power_law_1.2,2.2680192947387696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,8,1,power_law_1.2,0.7393407821655273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,8,1,power_law_1.01,0.5856319904327393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,8,1,power_law_1.2,0.3952575922012329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,8,1,power_law_1.2,0.7992191791534424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,8,1,power_law_1.2,0.42302718162536623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,8,1,power_law_1.01,0.7811391830444336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,8,1,power_law_1.2,0.08190079927444457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,8,1,power_law_1.2,0.928883171081543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,8,1,power_law_1.2,2.809612846374512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,8,1,power_law_1.2,0.43805441856384275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,8,1,power_law_1.2,0.11543040275573731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,8,1,power_law_1.01,3.6513729095458984
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,8,1,power_law_1.01,0.8646783828735352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,8,balanced,0.38526399930318195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,8,1,power_law_1.2,1.0562432289123536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,8,1,power_law_1.2,0.45543680191040037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,8,balanced,0.4254666566848755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,8,1,power_law_1.2,0.14376319646835328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,8,balanced,0.5013226668039957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,8,1,power_law_1.01,1.002950382232666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,8,1,power_law_1.2,0.5319488048553467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,8,balanced,0.578661322593689
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,8,1,power_law_1.01,1.1799488067626953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,8,1,power_law_1.2,0.17286399602890015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,8,1,power_law_1.2,1.318291187286377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,8,balanced,0.7867840131123861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,8,1,power_law_1.2,0.628652811050415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,8,1,power_law_1.2,0.17841919660568237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,8,1,power_law_1.01,1.2620608329772949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,8,balanced,0.9536426862080892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,8,1,power_law_1.2,0.18536959886550902
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,8,1,power_law_1.01,1.6744640350341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,8,1,power_law_1.01,4.708646392822265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,8,balanced,0.03573333223660787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,8,balanced,0.03565866748491923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,8,1,power_law_1.01,1.524614429473877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,8,1,power_law_1.2,0.19283839464187622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,8,balanced,0.04794133206208547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,8,balanced,0.05177066723505656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,8,balanced,0.060602664947509766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,8,balanced,0.06850133339564006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,8,1,power_law_1.2,0.2046207904815674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,8,1,power_law_1.2,0.21377279758453369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,8,1,power_law_1.01,2.0260223388671874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,8,1,power_law_1.2,0.23219199180603028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,8,1,power_law_1.2,3.860460662841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,8,1,power_law_1.2,0.25087358951568606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,8,1,power_law_1.2,0.775987195968628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,8,balanced,1.3424053192138672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,8,1,power_law_1.2,0.26737279891967775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,8,balanced,0.043920000394185386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,8,balanced,0.06897066533565521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,8,balanced,0.09845333298047383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,8,1,power_law_1.2,0.9632127761840821
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,8,1,power_law_1.2,0.30078079700469973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,8,balanced,0.1467413306236267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,8,balanced,0.1011199951171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,8,balanced,0.178874671459198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,8,balanced,0.17961066961288452
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,8,1,power_law_1.2,0.38059520721435547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,8,balanced,0.1827039917310079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,8,1,power_law_1.2,1.0447936058044434
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,8,1,power_law_1.2,0.3440576076507568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,8,1,power_law_1.2,5.05164794921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,8,1,power_law_1.2,0.4364352226257324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,8,1,power_law_1.2,1.307692813873291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,8,1,power_law_1.2,1.5712384223937987
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,8,1,power_law_1.2,0.4288191795349121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,8,1,power_law_1.01,2.5313343048095702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,8,balanced,1.720479965209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,8,balanced,0.2193173368771871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,8,balanced,0.22023467222849527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,8,1,power_law_1.2,2.0905536651611327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,8,balanced,0.18518400192260742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,8,balanced,2.619445323944092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,8,balanced,0.1874986688296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,8,balanced,0.1940000057220459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,8,balanced,0.19931199153264365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,8,1,power_law_1.01,3.5472320556640624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,8,balanced,0.2034346659978231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,8,balanced,0.21582933266957602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,8,balanced,0.22664533058802286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,8,balanced,0.23811733722686768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,8,balanced,0.2590026656786601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,8,balanced,3.540090560913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,8,balanced,0.3025226593017578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,8,balanced,0.34344534079233807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,8,power_law_1.01,0.0671999990940094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,8,balanced,0.02752533306678136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,8,balanced,0.42369600137074787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,8,1,power_law_1.2,1.5537728309631347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,8,power_law_1.01,0.1000704050064087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,8,balanced,0.509605328241984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,8,balanced,0.22427199284235635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,8,balanced,0.2260106603304545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,8,balanced,0.7167200247446696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,8,power_law_1.01,0.13521920442581176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,8,balanced,0.23157866795857748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,8,balanced,0.23694399992624918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,8,balanced,0.8783840338389078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,8,power_law_1.01,0.14090880155563354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,8,1,power_law_1.2,2.064563179016113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,8,balanced,1.251413345336914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,8,1,power_law_1.2,2.622208023071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,8,balanced,1.6207200686136882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,8,1,power_law_1.2,0.5577600002288818
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,8,balanced,0.027935999135176342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,8,1,power_law_1.2,2.562073516845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,8,balanced,0.03038399914900462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,8,balanced,0.03345600018898646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,8,balanced,0.04970133304595947
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,8,1,power_law_1.2,0.6402624130249024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,8,balanced,0.07042133311430614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,8,balanced,0.07259733478228252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,8,balanced,0.07492800056934357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,8,1,power_law_1.01,4.546220779418945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,8,balanced,0.241376002629598
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,8,1,power_law_1.2,0.7963903903961181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,8,balanced,0.25496000051498413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,8,balanced,0.2653226653734843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,8,power_law_1.01,0.15676800012588502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,8,balanced,0.27349332968393963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,8,balanced,0.2964479923248291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,8,1,power_law_1.2,3.571865463256836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,8,balanced,2.358736038208008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,8,balanced,0.07453866799672444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,8,balanced,0.07419200241565704
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,8,balanced,0.07543466488520305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,8,balanced,0.0749013324578603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,8,power_law_1.01,0.06092159748077393
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,8,balanced,0.07638399799664815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,8,balanced,3.0954507191975913
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,8,balanced,0.07746666669845581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,8,1,power_law_1.2,4.575276947021484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,8,balanced,0.10362666845321655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,8,power_law_1.01,0.09481599926948547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,8,balanced,0.09955733021100362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,8,power_law_1.01,0.1737215995788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,8,power_law_1.01,0.060089600086212155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,8,power_law_1.01,0.10117759704589843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,8,power_law_1.01,0.22639999389648438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,8,power_law_1.01,0.09283199906349182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,8,power_law_1.01,0.11442559957504272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,8,power_law_1.01,0.25902719497680665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,8,power_law_1.01,0.10117759704589843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,8,power_law_1.01,0.12485120296478272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,8,power_law_1.01,0.24353280067443847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,8,power_law_1.01,0.11260160207748413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,8,power_law_1.01,0.13070080280303956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,8,1,power_law_1.2,3.677964782714844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,8,power_law_1.01,0.26467840671539306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,8,power_law_1.01,0.12726399898529053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,8,power_law_1.01,0.20626559257507324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,8,power_law_1.01,0.2695359945297241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,8,power_law_1.01,0.1518847942352295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,8,power_law_1.01,0.20742399692535402
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,8,1,power_law_1.2,0.882265567779541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,8,power_law_1.01,0.17392640113830565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,8,power_law_1.01,0.20874879360198975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,8,power_law_1.01,0.1674239993095398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,8,power_law_1.01,0.22263040542602539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,8,1,power_law_1.2,1.1964672088623047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,8,power_law_1.01,0.1686336040496826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,8,power_law_1.01,0.22432000637054444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,8,1,power_law_1.2,4.720390319824219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,8,power_law_1.01,0.0286655992269516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,8,power_law_1.01,0.23021440505981444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,8,power_law_1.2,0.06817920207977295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,8,power_law_1.01,0.23467519283294677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,8,1,power_law_1.2,1.6898239135742188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,8,power_law_1.2,0.09733759760856628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,8,power_law_1.01,0.2721152067184448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,8,balanced,0.3386400143305461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,8,power_law_1.2,0.06072319746017456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,8,power_law_1.2,0.11400959491729737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,8,balanced,0.38011733690897626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,8,power_law_1.01,0.28190720081329346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,8,power_law_1.2,0.09408640265464782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,8,power_law_1.2,0.1460736036300659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,8,power_law_1.01,0.2919872045516968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,8,power_law_1.2,0.10076160430908203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,8,power_law_1.2,0.17271679639816284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,8,power_law_1.01,0.30487680435180664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,8,power_law_1.2,0.106278395652771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,8,power_law_1.01,0.31480960845947265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,8,power_law_1.01,0.17032959461212158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,8,power_law_1.2,0.12333439588546753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,8,power_law_1.01,0.3315711975097656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,8,power_law_1.01,0.1877120018005371
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,8,power_law_1.01,0.044998401403427125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,8,power_law_1.2,0.11904640197753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,8,power_law_1.01,0.37795200347900393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,8,power_law_1.01,0.2485759973526001
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,8,power_law_1.01,0.05448319911956787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,8,power_law_1.2,0.20100479125976561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,8,power_law_1.01,0.4197120189666748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,8,power_law_1.01,0.26131839752197267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,8,power_law_1.01,0.05114240050315857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,8,power_law_1.2,0.19708160161972046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,8,balanced,0.45907731850941974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,8,power_law_1.01,0.2714495897293091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,8,power_law_1.01,0.5490816116333008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,8,power_law_1.01,0.05741440057754517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,8,power_law_1.2,0.21496319770812988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,8,power_law_1.01,0.29731199741363523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,8,power_law_1.01,0.6515647888183593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,8,power_law_1.01,0.0639743983745575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,8,power_law_1.2,0.20920319557189943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,8,power_law_1.01,0.34932479858398435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,8,power_law_1.2,0.1461632013320923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,8,power_law_1.2,0.2215359926223755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,8,power_law_1.01,0.8770624160766601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,8,power_law_1.01,0.4060671806335449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,8,power_law_1.2,0.23006720542907716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,8,power_law_1.01,0.1858240008354187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,8,power_law_1.01,0.5070079803466797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,8,power_law_1.2,0.2529344081878662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,8,power_law_1.01,0.18851840496063232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,8,power_law_1.2,0.23882880210876464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,8,power_law_1.01,0.20633599758148194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,8,power_law_1.2,0.24801280498504638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,8,power_law_1.01,0.21548800468444823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,8,power_law_1.2,0.26023039817810056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,8,power_law_1.01,0.23067519664764405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,8,power_law_1.2,0.263155198097229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,8,power_law_1.01,0.26071040630340575
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,8,power_law_1.01,0.06478080153465271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,8,power_law_1.2,0.2812608003616333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,8,power_law_1.01,0.3091840028762817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,8,power_law_1.2,0.2289344072341919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,8,power_law_1.01,0.06736000180244446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,8,power_law_1.2,0.29506559371948243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,8,power_law_1.2,0.23626880645751952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,8,power_law_1.01,1.087775993347168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,8,power_law_1.2,0.30576000213623045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,8,balanced,0.0902453362941742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,8,power_law_1.2,0.25144960880279543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,8,balanced,0.1684053341547648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,8,power_law_1.01,0.6192895889282226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,8,balanced,0.12158933281898499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,8,power_law_1.2,0.2632704019546509
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,8,balanced,0.10930132865905762
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,8,balanced,0.1197653313477834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,8,balanced,0.5442186594009399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,8,power_law_1.2,0.27382400035858157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,8,power_law_1.01,0.8661631584167481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,8,balanced,0.7632799943288168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,8,power_law_1.2,0.299238395690918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,8,balanced,0.9326346715291342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,8,power_law_1.01,1.0770432472229003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,8,power_law_1.2,0.35306239128112793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,8,balanced,1.3237333297729492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,8,power_law_1.01,0.3607615947723389
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,8,power_law_1.01,0.06746879816055298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,8,power_law_1.01,1.5094207763671874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,8,balanced,1.7141653696695964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,8,power_law_1.01,0.4688576221466064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,8,power_law_1.2,0.31067519187927245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,8,power_law_1.01,1.506726360321045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,8,power_law_1.01,0.5586751937866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,8,balanced,0.1395146648089091
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,8,balanced,0.16211199760437012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,8,power_law_1.01,1.938528060913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,8,balanced,2.4982239405314126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,8,balanced,0.22802666823069254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,8,power_law_1.01,0.7670207977294922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,8,balanced,0.2737013300259908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,8,balanced,0.3879306713740031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,8,power_law_1.01,1.9962495803833007
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,8,balanced,0.5023306608200073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,8,power_law_1.01,0.9627455711364746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,8,balanced,3.282479921976725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,8,power_law_1.01,2.9349695205688477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,8,power_law_1.2,0.41864957809448244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,8,power_law_1.01,0.06673920154571533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,8,power_law_1.01,1.3781760215759278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,8,power_law_1.01,2.885312080383301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,8,power_law_1.2,0.5361023902893066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,8,power_law_1.01,1.8295232772827148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,8,power_law_1.2,0.646073579788208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,8,power_law_1.01,3.815398406982422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,8,1,balanced,0.025424001117547352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,8,1,balanced,0.03206400076548258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,8,1,balanced,0.04722133278846741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,8,1,balanced,0.06494933366775513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,8,1,balanced,0.10802666346232097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,8,power_law_1.01,3.7000961303710938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,8,1,balanced,0.1955946683883667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,8,1,balanced,0.029440000653266907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,8,1,balanced,0.1949440042177836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,8,1,balanced,0.03339733431736628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,8,1,balanced,0.19591999053955078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,8,1,balanced,0.1973653237024943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,8,1,balanced,0.19730132818222046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,8,power_law_1.01,0.07025920152664185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,8,1,balanced,0.19684267044067383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,8,1,balanced,0.19904534022013345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,8,1,balanced,0.20257065693537393
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,8,power_law_1.01,0.06865280270576476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,8,1,balanced,0.20758400360743204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,8,1,balanced,0.24989332755406699
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,8,power_law_1.01,0.06891520023345947
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,8,power_law_1.2,0.029875200986862183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,8,power_law_1.01,0.07260800004005433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,8,power_law_1.2,0.04509440064430237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,8,power_law_1.01,0.08039039969444275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,8,power_law_1.2,0.04498560130596161
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,8,power_law_1.01,0.08021759986877441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,8,power_law_1.2,0.060627198219299315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,8,power_law_1.01,2.6519359588623046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,8,power_law_1.01,0.08162559866905213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,8,1,balanced,0.041706666350364685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,8,1,balanced,0.0580320010582606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,8,1,balanced,0.10318932930628459
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,8,power_law_1.01,0.1182144045829773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,8,power_law_1.2,0.33125760555267336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,8,power_law_1.01,0.11543680429458618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,8,power_law_1.2,0.38454399108886717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,8,power_law_1.01,0.11265280246734619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,8,1,balanced,0.2552799979845683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,8,power_law_1.2,0.4524223804473877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,8,power_law_1.01,0.12817920446395875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,8,power_law_1.01,3.4262977600097657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,8,power_law_1.2,0.8933247566223145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,8,power_law_1.2,0.5736576080322265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,8,power_law_1.2,0.6869823932647705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,8,power_law_1.2,1.168006420135498
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,8,power_law_1.2,0.05115519762039185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,8,power_law_1.2,0.09288960099220275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,8,1,balanced,0.16319466630617777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,8,power_law_1.2,0.9031807899475097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,8,power_law_1.2,0.09629439711570739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,8,power_law_1.2,1.636307144165039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,8,power_law_1.2,1.1262528419494628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,8,1,balanced,0.26383999983469647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,8,1,balanced,0.27830400069554645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,8,1,balanced,0.29609066247940063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,8,1,balanced,0.3300373355547587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,8,power_law_1.2,2.063199996948242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,8,power_law_1.2,1.6249343872070312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,8,1,balanced,0.025445332129796345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,8,1,balanced,0.027066667874654133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,8,1,balanced,0.02945599953333537
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,8,1,balanced,0.03398400048414866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,8,power_law_1.2,0.057625597715377806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,8,power_law_1.2,2.1060800552368164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,8,power_law_1.2,0.06449919939041138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,8,power_law_1.2,2.9691648483276367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,8,power_law_1.2,0.06487680077552796
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,8,power_law_1.01,0.15670399665832518
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,8,power_law_1.2,0.06765440106391907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,8,1,balanced,0.3648159901301066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,8,power_law_1.2,3.116979217529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,8,power_law_1.2,0.0679423987865448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,8,1,balanced,0.48427732785542804
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,8,power_law_1.2,0.06622080206871032
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,8,power_law_1.2,0.06751360297203064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,8,1,balanced,0.05489600201447805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,8,1,balanced,0.07874133189519246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,8,1,balanced,0.08062933385372162
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,8,power_law_1.2,0.0716480016708374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,8,1,balanced,0.08271466692288716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,8,power_law_1.2,0.09943040013313294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,8,1,balanced,0.08338666955629985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,8,1,balanced,0.082805335521698
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,8,1,balanced,0.08682133754094441
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,8,1,balanced,0.08847999572753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,8,power_law_1.2,3.9812801361083983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,8,1,balanced,0.09282666444778442
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,8,1,balanced,0.09569600224494934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,8,1,balanced,0.10883733630180359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,8,power_law_1.01,0.19188480377197265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,8,power_law_1.01,0.25804159641265867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,8,power_law_1.2,4.020025634765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,8,1,balanced,0.16499732931454977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,8,power_law_1.01,0.3278271913528442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,8,1,balanced,0.16366400321324667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,8,1,balanced,0.16556266943613687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,8,1,power_law_1.01,0.02561280131340027
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,8,power_law_1.01,0.4690239906311035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,8,power_law_1.2,0.07171199917793274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,8,1,power_law_1.01,0.030585598945617676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,8,power_law_1.2,0.13151999711990356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,8,power_law_1.2,0.07392640113830566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,8,1,power_law_1.01,0.043750399351119997
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,8,power_law_1.01,0.6844160079956054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,8,power_law_1.2,0.16354559659957885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,8,power_law_1.2,0.07936639785766601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,8,1,power_law_1.01,0.022777600586414336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,8,1,balanced,0.11212799946467082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,8,power_law_1.2,0.08697599768638611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,8,power_law_1.2,0.1621888041496277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,8,1,power_law_1.01,0.027871999144554137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,8,power_law_1.2,0.08609279990196228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,8,power_law_1.2,0.16963839530944824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,8,1,power_law_1.01,0.046419200301170346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,8,power_law_1.2,0.14005759954452515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,8,power_law_1.2,0.16188160181045533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,8,1,power_law_1.01,0.063372802734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,8,power_law_1.2,0.11085439920425415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,8,1,balanced,0.16506666938463846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,8,1,power_law_1.01,0.08863999843597412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,8,1,balanced,0.1655946671962738
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,8,power_law_1.2,0.1146239995956421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,8,1,power_law_1.01,0.11233279705047608
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,8,power_law_1.2,0.13303040266036986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,8,1,power_law_1.01,0.15527679920196533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,8,1,power_law_1.01,0.05804799795150757
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,8,power_law_1.2,0.1655295968055725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,8,1,power_law_1.01,0.16709760427474976
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,8,1,balanced,0.12324800093968709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,8,1,power_law_1.01,0.08213760256767273
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,8,1,balanced,0.12586667140324911
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,8,power_law_1.2,0.20356481075286864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,8,1,power_law_1.01,0.10851199626922607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,8,power_law_1.2,0.2748863935470581
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,8,1,power_law_1.01,0.023443199694156647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,8,1,power_law_1.01,0.12766079902648925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,8,power_law_1.2,0.17354240417480468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,8,1,power_law_1.01,0.13226879835128785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,8,1,balanced,0.16894400119781494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,8,1,balanced,0.1715679963429769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,8,power_law_1.2,0.1831168055534363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,8,1,balanced,0.17549866437911987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,8,1,balanced,0.20238399505615234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,8,power_law_1.2,0.18316160440444945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,8,1,balanced,0.20511466264724731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,8,1,balanced,0.21612266699473062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,8,1,balanced,0.23682133356730142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,8,power_law_1.2,0.19247360229492189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,8,1,balanced,0.25430933634440106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,8,1,balanced,0.29333333174387616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,8,1,power_law_1.01,0.17036800384521483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,8,1,balanced,0.1381386617819468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,8,1,balanced,0.33268799384435016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,8,1,balanced,0.14803199966748556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,8,1,balanced,0.46724800268809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,8,1,balanced,0.5480320056279501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,8,1,balanced,0.7684480349222819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,8,1,power_law_1.01,0.13379199504852296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,8,1,power_law_1.01,0.1370751976966858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,8,1,power_law_1.01,0.1428928017616272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,8,1,power_law_1.01,0.15387519598007202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,8,1,balanced,0.5594613154729208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,8,1,power_law_1.01,0.163046395778656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,8,1,balanced,0.1667893330256144
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,8,1,balanced,0.24025599161783853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,8,1,power_law_1.01,0.1831936001777649
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,8,1,balanced,0.24901332457860312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,8,1,balanced,0.418938676516215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,8,1,power_law_1.01,0.19682559967041016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,8,1,balanced,0.3696800072987874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,8,1,balanced,0.982917308807373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,8,1,power_law_1.01,0.2034048080444336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,8,1,balanced,0.520090659459432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,8,1,power_law_1.01,0.2131648063659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,8,1,balanced,0.6711200078328451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,8,1,balanced,1.424789269765218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,8,1,power_law_1.01,0.24002559185028077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,8,1,power_law_1.2,0.022592000663280487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,8,power_law_1.2,0.2056191921234131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,8,1,power_law_1.01,0.26406400203704833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,8,1,power_law_1.2,0.02910720109939575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,8,1,balanced,1.8638879458109539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,8,power_law_1.2,0.22227199077606202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,8,1,power_law_1.01,0.1760640025138855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,8,1,power_law_1.2,0.025254398584365845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,8,power_law_1.2,0.23519999980926515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,8,1,power_law_1.01,0.18246400356292725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,8,1,power_law_1.2,0.03059839904308319
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,8,power_law_1.2,0.34524800777435305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,8,power_law_1.2,0.2605247974395752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,8,1,power_law_1.01,0.18357759714126587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,8,1,power_law_1.2,0.044940799474716187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,8,power_law_1.2,0.3151616096496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,8,1,power_law_1.2,0.058143997192382814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,8,power_law_1.2,0.364134407043457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,8,1,power_law_1.2,0.07492480278015137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,8,1,power_law_1.2,0.0974399983882904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,8,1,power_law_1.2,0.12634880542755128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,8,1,balanced,0.7789973417917887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,8,1,power_law_1.2,0.13141759634017944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,8,1,balanced,0.9906506538391113
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,8,1,power_law_1.01,0.024300800263881685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,8,1,power_law_1.2,0.13388799428939818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,8,1,balanced,1.424239953358968
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,8,1,power_law_1.01,0.026950401067733765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,8,1,power_law_1.01,0.18729599714279174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,8,1,power_law_1.01,0.20223360061645507
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,8,1,power_law_1.01,0.03467519879341126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,8,1,balanced,1.852128028869629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,8,1,power_law_1.01,0.24143359661102295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,8,1,power_law_1.01,0.04490880072116852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,8,power_law_1.2,0.4693888187408447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,8,1,power_law_1.01,0.24515841007232667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,8,1,power_law_1.01,0.060127997398376466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,8,1,power_law_1.01,0.32039039134979247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,8,power_law_1.2,0.586297607421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,8,1,power_law_1.01,0.06883199810981751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,8,1,power_law_1.01,0.37136640548706057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,8,power_law_1.2,0.8165696144104004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,8,1,power_law_1.01,0.4848959922790527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,8,1,power_law_1.2,0.13827840089797974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,8,power_law_1.2,0.9824959754943847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,8,1,power_law_1.2,0.1464959979057312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,8,1,power_law_1.01,0.5871744155883789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,8,1,power_law_1.2,0.15362559556961058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,8,1,power_law_1.01,0.809382438659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,8,power_law_1.2,1.4214207649230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,8,1,power_law_1.2,0.04694400131702423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,8,1,power_law_1.01,1.0271552085876465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,8,1,power_law_1.2,0.062003201246261595
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,8,1,power_law_1.2,0.02535040080547333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,8,1,power_law_1.01,0.07159680128097534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,8,1,power_law_1.2,0.08224639892578126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,8,1,power_law_1.2,0.02563839852809906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,8,1,power_law_1.01,0.07345280051231384
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,8,power_law_1.2,0.48847360610961915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,8,1,power_law_1.2,0.10430719852447509
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,8,1,power_law_1.2,0.028070399165153505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,8,1,power_law_1.01,0.07728639841079712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,8,1,power_law_1.2,0.1580415964126587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,8,1,power_law_1.2,0.03487359881401062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,8,power_law_1.2,0.7586112022399902
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,8,1,power_law_1.01,0.08024320006370544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,8,balanced,0.03346133232116699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,8,1,power_law_1.2,0.17187199592590333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,8,1,power_law_1.2,0.1665536046028137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,8,1,power_law_1.01,0.08303359746932984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,8,1,power_law_1.2,0.17464959621429443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,8,1,power_law_1.01,0.25438079833984373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,8,1,power_law_1.01,0.0868607997894287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,8,1,power_law_1.2,0.178438401222229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,8,1,power_law_1.01,0.2773632049560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,8,1,power_law_1.01,0.09249920248985291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,8,power_law_1.2,1.8556991577148438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,8,1,power_law_1.2,0.18017280101776123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,8,1,power_law_1.01,0.10064640045166015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,8,1,power_law_1.2,0.18214399814605714
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,8,1,power_law_1.01,0.1072767972946167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,8,1,power_law_1.2,0.1866943955421448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,8,1,power_law_1.01,0.12332160472869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,8,1,power_law_1.2,0.20551040172576904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,8,1,power_law_1.01,0.12006399631500245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,8,power_law_1.2,2.7128000259399414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,8,1,power_law_1.2,0.24083199501037597
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,8,1,power_law_1.01,0.14019839763641356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,8,1,power_law_1.2,0.18499840497970582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,8,1,power_law_1.01,0.14640640020370482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,8,1,power_law_1.2,0.1981503963470459
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,8,1,power_law_1.01,0.1713152050971985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,8,1,power_law_1.2,0.20404479503631592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,8,1,power_law_1.01,0.30439679622650145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,8,1,power_law_1.01,0.23177599906921387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,8,1,power_law_1.2,0.21620481014251708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,8,1,power_law_1.01,1.4667327880859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,8,1,power_law_1.01,0.268339204788208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,8,1,power_law_1.2,0.2434688091278076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,8,power_law_1.2,3.762054443359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,8,balanced,0.03551999976237615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,8,balanced,0.03551466763019562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,8,balanced,0.04459733267625173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,8,balanced,0.06614399949709575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,8,balanced,0.0979306697845459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,8,balanced,0.035818666219711304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,8,1,power_law_1.2,0.04270080029964447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,8,balanced,0.05691199998060862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,8,1,power_law_1.01,1.9084543228149413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,8,balanced,0.07306133210659027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,8,balanced,0.10771200060844421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,8,1,power_law_1.2,0.24556798934936525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,8,balanced,0.18941332896550497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,8,balanced,0.04688533147176107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,8,balanced,0.2627573410669963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,8,balanced,0.05201066533724467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,8,1,power_law_1.2,0.25470719337463377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,8,balanced,0.05641066531340281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,8,balanced,0.2648586630821228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,8,balanced,0.0681386689345042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,8,balanced,0.2690933346748352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,8,1,power_law_1.2,0.27372159957885744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,8,1,power_law_1.2,0.3272511959075928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,8,1,power_law_1.01,0.36154880523681643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,8,1,power_law_1.2,0.3640768051147461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,8,1,power_law_1.01,0.32441599369049073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,8,1,power_law_1.2,0.406771183013916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,8,balanced,0.14498666922251383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,8,balanced,0.21832533677419028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,8,balanced,0.22166399161020914
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,8,1,power_law_1.2,0.05883520245552063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,8,balanced,0.2241119941075643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,8,balanced,0.22523200511932373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,8,balanced,0.23200533787409464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,8,1,power_law_1.2,0.06867200136184692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,8,balanced,0.10244799653689067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,8,balanced,0.1768959959348043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,8,1,power_law_1.2,0.07166079878807068
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,8,1,power_law_1.2,0.07489920258522034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,8,1,power_law_1.01,0.40341758728027344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,8,1,power_law_1.2,0.07612159848213196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,8,1,power_law_1.2,0.27381119728088377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,8,1,power_law_1.2,0.08095999956130981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,8,1,power_law_1.01,0.5089087963104248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,8,1,power_law_1.2,0.3294912099838257
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,8,1,power_law_1.01,0.3986304044723511
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,8,1,power_law_1.2,0.08421760201454162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,8,1,power_law_1.2,0.5208960056304932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,8,1,power_law_1.2,0.3807039976119995
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,8,1,power_law_1.2,0.08759679794311523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,8,1,power_law_1.01,0.5458879947662354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,8,1,power_law_1.2,0.6133567810058593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,8,balanced,0.23847466707229614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,8,balanced,0.2692799965540568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,8,balanced,0.24055467049280801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,8,balanced,0.1813653310139974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,8,balanced,0.25461333990097046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,8,balanced,0.18129066626230875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,8,1,power_law_1.2,0.8359871864318847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,8,balanced,0.18592000007629395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,8,balanced,0.1890666683514913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,8,balanced,0.19339734315872192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,8,balanced,0.19940267006556192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,8,balanced,0.20316267013549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,8,balanced,0.21585599581400552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,8,balanced,0.22707732518513998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,8,balanced,0.23799467086791992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,8,balanced,0.2598399917284648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,8,1,power_law_1.01,0.6062272071838379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,8,balanced,0.3032640020052592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,8,balanced,0.3418026765187581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,8,1,power_law_1.2,0.49031682014465333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,8,balanced,0.4222986698150635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,8,balanced,0.5089333454767863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,8,1,power_law_1.2,0.5980288028717041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,8,balanced,0.26498667399088544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,8,balanced,0.7137973308563232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,8,balanced,0.27535466353098553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,8,balanced,0.29789332548777264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,8,balanced,0.8752319812774658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,8,balanced,0.3389279842376709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,8,1,power_law_1.2,1.053337574005127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,8,balanced,0.380570650100708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,8,balanced,0.46140801906585693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,8,balanced,0.546554684638977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,8,1,power_law_1.2,1.4881279945373536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,8,balanced,0.7645653088887533
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,8,1,power_law_1.2,0.09464960098266602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,8,1,power_law_1.01,0.8172479629516601
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,8,1,power_law_1.01,0.7839807987213134
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,8,1,power_law_1.2,0.10175360441207885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,8,1,power_law_1.2,0.10823040008544922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,8,1,power_law_1.2,1.9327104568481446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,8,1,power_law_1.2,0.8168000221252442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,8,balanced,1.2433599630991619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,8,power_law_1.01,0.06789759993553161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,8,1,power_law_1.2,1.0363007545471192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,8,power_law_1.01,0.0992959976196289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,8,balanced,1.6164906819661458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,8,power_law_1.01,0.12494720220565796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,8,balanced,0.9351092974344889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,8,balanced,0.27432000637054443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,8,power_law_1.01,0.14385279417037963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,8,balanced,0.28119999170303345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,8,balanced,1.3287359873453777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,8,balanced,2.3504586219787598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,8,1,power_law_1.01,1.041215991973877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,8,power_law_1.01,0.17593599557876588
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,8,1,power_law_1.2,0.1260607957839966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,8,balanced,1.7187093098958333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,8,power_law_1.01,0.1843008041381836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,8,1,power_law_1.2,0.12176640033721924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,8,1,power_law_1.01,1.4714752197265626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,8,power_law_1.01,0.2512768030166626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,8,balanced,3.087285359700521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,8,1,power_law_1.2,0.1412287950515747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,8,balanced,2.504864056905111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,8,power_law_1.01,0.25518720149993895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,8,power_law_1.01,0.06087679862976074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,8,1,power_law_1.2,0.14676480293273925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,8,power_law_1.01,0.25975680351257324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,8,power_law_1.01,0.09372159838676453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,8,1,power_law_1.01,1.9041408538818358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,8,1,power_law_1.2,1.4847423553466796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,8,power_law_1.01,0.24903039932250975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,8,power_law_1.01,0.09886720180511474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,8,balanced,0.28700800736745197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,8,balanced,3.2893012364705405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,8,power_law_1.01,0.26873600482940674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,8,power_law_1.01,0.061222398281097413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,8,balanced,0.29850133260091144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,8,balanced,0.3099626700083415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,8,balanced,0.31963199377059937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,8,power_law_1.01,0.09447680115699768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,8,balanced,0.34225066502888996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,8,balanced,0.384768009185791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,8,balanced,0.02552533398071925
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,8,balanced,0.02756800005833308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,8,balanced,0.42529598871866864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,8,balanced,0.027776000400384266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,8,balanced,0.03366933266321818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,8,balanced,0.5016053517659506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,8,balanced,0.5781386693318685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,8,1,power_law_1.2,0.17320319414138793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,8,balanced,0.7879520257314047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,8,1,power_law_1.2,0.22104959487915038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,8,balanced,0.9523999691009521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,8,1,power_law_1.2,0.27054719924926757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,8,power_law_1.01,0.27796480655670164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,8,1,power_law_1.2,0.32549760341644285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,8,balanced,1.3400586446126301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,8,1,power_law_1.2,1.9304512023925782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,8,power_law_1.01,0.275052809715271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,8,power_law_1.2,0.06821119785308838
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,8,1,power_law_1.2,0.4058432102203369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,8,power_law_1.01,0.29407999515533445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,8,balanced,1.7173280715942383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,8,power_law_1.2,0.09651839733123779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,8,power_law_1.01,0.3053312063217163
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,8,1,power_law_1.2,0.5521599769592285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,8,power_law_1.2,0.09676160216331482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,8,power_law_1.01,0.31089279651641843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,8,power_law_1.2,0.1388159990310669
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,8,1,power_law_1.2,0.776416015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,8,power_law_1.01,0.3271615982055664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,8,power_law_1.2,0.16539520025253296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,8,power_law_1.01,0.10440959930419921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,8,power_law_1.01,0.3755264043807983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,8,power_law_1.2,0.11385600566864014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,8,power_law_1.01,0.11539839506149292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,8,power_law_1.01,0.43080959320068357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,8,power_law_1.2,0.2520960092544556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,8,power_law_1.01,0.13576960563659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,8,power_law_1.01,0.5459328174591065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,8,power_law_1.01,0.10947840213775635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,8,power_law_1.2,0.2518399953842163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,8,power_law_1.01,0.20778241157531738
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,8,power_law_1.01,0.030220800638198854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,8,power_law_1.01,0.6474495887756347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,8,power_law_1.2,0.24714879989624022
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,8,power_law_1.01,0.046348801255226134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,8,power_law_1.2,0.26133759021759034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,8,power_law_1.01,0.8729087829589843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,8,power_law_1.01,0.05180799961090088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,8,power_law_1.2,0.27311999797821046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,8,power_law_1.2,0.06120960116386413
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,8,power_law_1.01,0.05730559825897217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,8,power_law_1.01,1.1096832275390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,8,power_law_1.2,0.26513280868530276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,8,power_law_1.2,0.09219200015068055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,8,power_law_1.01,0.05989760160446167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,8,power_law_1.2,0.2793152093887329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,8,power_law_1.2,0.09123839735984803
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,8,power_law_1.01,0.06551679968833923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,8,power_law_1.01,1.5354623794555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,8,power_law_1.01,0.06282879710197449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,8,power_law_1.01,0.10760960578918458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,8,power_law_1.01,0.2124608039855957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,8,power_law_1.2,0.10160000324249267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,8,power_law_1.01,0.06848639845848084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,8,balanced,2.6070186297098794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,8,power_law_1.01,0.21692159175872802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,8,power_law_1.2,0.11886719465255738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,8,power_law_1.01,0.22099199295043945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,8,power_law_1.2,0.10639359951019287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,8,power_law_1.01,0.22668159008026123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,8,power_law_1.2,0.21420159339904785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,8,power_law_1.01,0.23175039291381835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,8,balanced,3.5549119313557944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,8,power_law_1.2,0.2103424072265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,8,power_law_1.2,0.2928960084915161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,8,power_law_1.2,0.20421760082244872
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,8,balanced,0.05050133168697357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,8,power_law_1.2,0.060102397203445436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,8,power_law_1.2,0.3022655963897705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,8,balanced,0.06921599805355072
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,8,balanced,0.07259733478228252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,8,balanced,0.07460266848405202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,8,balanced,0.07314133147398631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,8,power_law_1.01,0.1368191957473755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,8,balanced,0.07372800012429555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,8,power_law_1.01,2.0909120559692385
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,8,power_law_1.01,0.06714239716529846
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,8,balanced,0.0749120016892751
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,8,balanced,0.07417599856853485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,8,power_law_1.01,0.16579840183258057
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,8,balanced,0.0764213353395462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,8,power_law_1.01,0.06830080151557923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,8,balanced,0.07707199951012929
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,8,balanced,0.10390933354695638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,8,power_law_1.01,0.07030400037765502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,8,power_law_1.01,0.1646016001701355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,8,power_law_1.01,0.06919040083885193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,8,power_law_1.01,0.16747519969940186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,8,power_law_1.01,2.812371253967285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,8,power_law_1.01,0.23839359283447265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,8,power_law_1.01,0.16860159635543823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,8,power_law_1.01,0.25032958984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,8,power_law_1.01,0.17282559871673583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,8,power_law_1.2,0.21604480743408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,8,power_law_1.01,0.2633599996566772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,8,power_law_1.2,0.09300479888916016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,8,power_law_1.01,0.1778496026992798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,8,power_law_1.2,0.22643840312957764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,8,power_law_1.2,0.09498879909515381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,8,power_law_1.01,0.18762880563735962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,8,power_law_1.2,0.23183999061584473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,8,power_law_1.2,0.10061440467834473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,8,power_law_1.01,0.1982591986656189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,8,power_law_1.2,0.23888640403747557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,8,power_law_1.01,3.8339775085449217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,8,balanced,0.10130133231480916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,8,balanced,0.08934932947158813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,8,power_law_1.2,0.12112640142440796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,8,power_law_1.2,0.25011839866638186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,8,power_law_1.2,0.2614464044570923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,8,power_law_1.01,0.06920959949493408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,8,power_law_1.2,0.27445120811462403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,8,power_law_1.2,0.29845120906829836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,8,power_law_1.2,0.30856320858001707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,8,power_law_1.2,0.3488640069961548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,8,power_law_1.01,0.2726399898529053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,8,power_law_1.2,0.41989760398864745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,8,power_law_1.01,0.2993216037750244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,8,power_law_1.01,0.3513727903366089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,8,power_law_1.2,0.525600004196167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,8,balanced,0.16710400581359863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,8,balanced,0.12682666381200156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,8,balanced,0.11016533772150676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,8,power_law_1.01,0.39719040393829347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,8,balanced,0.1207413375377655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,8,power_law_1.2,0.6494527816772461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,8,balanced,0.14075199762980142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,8,power_law_1.01,0.07129600048065185
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,8,balanced,0.16196800271670023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,8,power_law_1.01,0.5150400161743164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,8,balanced,0.2283359964688619
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,8,power_law_1.01,0.08560640215873719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,8,balanced,0.2755039930343628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,8,power_law_1.2,0.028998398780822755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,8,power_law_1.01,0.07863680124282837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,8,balanced,0.3887573480606079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,8,power_law_1.2,0.3283263921737671
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,8,balanced,0.5041973193486532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,8,1,balanced,0.023306667804718018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,8,1,balanced,0.03664533297220866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,8,power_law_1.2,0.38511359691619873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,8,power_law_1.01,0.2032639980316162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,8,power_law_1.2,0.45012478828430175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,8,power_law_1.01,0.21866240501403808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,8,power_law_1.2,0.165555202960968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,8,power_law_1.2,0.5647552013397217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,8,power_law_1.01,0.22612481117248534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,8,power_law_1.2,0.17568000555038452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,8,power_law_1.01,0.2558784008026123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,8,power_law_1.2,0.685478401184082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,8,power_law_1.2,0.1673151969909668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,8,power_law_1.01,0.6314815998077392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,8,power_law_1.2,0.04509440064430237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,8,power_law_1.2,0.17987200021743774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,8,power_law_1.2,0.9019007682800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,8,power_law_1.2,0.04480000138282776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,8,power_law_1.2,0.16877440214157105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,8,power_law_1.2,0.05253120064735413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,8,power_law_1.2,1.1587008476257323
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,8,power_law_1.2,0.05947520136833191
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,8,power_law_1.2,0.05800319910049438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,8,power_law_1.2,1.6239423751831055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,8,power_law_1.2,0.8748543739318848
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,8,power_law_1.2,0.06481919884681701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,8,power_law_1.01,0.30903680324554444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,8,power_law_1.2,0.06676480174064636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,8,power_law_1.01,0.0839680016040802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,8,power_law_1.01,0.3602816104888916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,8,power_law_1.2,1.1142144203186035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,8,power_law_1.2,2.0503871917724608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,8,power_law_1.01,0.8343104362487793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,8,power_law_1.01,0.11349760293960572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,8,power_law_1.01,0.45885438919067384
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,8,power_law_1.01,0.12057600021362305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,8,power_law_1.2,0.17800960540771485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,8,power_law_1.01,0.11244800090789794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,8,power_law_1.2,0.18444160223007203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,8,power_law_1.01,0.12665599584579468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,8,power_law_1.2,0.18853759765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,8,power_law_1.2,3.033401679992676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,8,power_law_1.2,0.20336639881134033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,8,power_law_1.2,0.22279040813446044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,8,power_law_1.2,0.229913592338562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,8,power_law_1.2,0.06290559768676758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,8,power_law_1.2,0.2635200023651123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,8,power_law_1.2,0.06636160016059875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,8,power_law_1.2,0.3219199895858765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,8,power_law_1.2,0.06655359864234925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,8,power_law_1.01,0.57260160446167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,8,power_law_1.2,1.5869952201843263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,8,power_law_1.2,0.36788480281829833
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,8,power_law_1.2,0.06776319742202759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,8,power_law_1.01,0.77226881980896
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,8,power_law_1.2,0.0709119975566864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,8,power_law_1.01,0.15821440219879152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,8,power_law_1.2,0.4765183925628662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,8,power_law_1.2,0.06977919936180114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,8,power_law_1.2,2.0044607162475585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,8,power_law_1.01,0.9693951606750488
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,8,power_law_1.2,0.0824512004852295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,8,power_law_1.2,0.08254079818725586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,8,1,balanced,0.047168001532554626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,8,1,balanced,0.06628799935181935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,8,1,balanced,0.10735999544461568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,8,power_law_1.01,1.3873984336853027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,8,power_law_1.2,3.9919807434082033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,8,1,balanced,0.19503466288248697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,8,1,balanced,0.029546665648619335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,8,1,balanced,0.19633066654205322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,8,1,balanced,0.03356266766786575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,8,1,balanced,0.19473065932591757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,8,power_law_1.01,1.052678394317627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,8,1,balanced,0.19570666551589966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,8,1,balanced,0.19672000408172607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,8,1,balanced,0.1970026691754659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,8,power_law_1.01,1.7763776779174805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,8,power_law_1.01,1.506009578704834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,8,power_law_1.01,0.1882688045501709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,8,power_law_1.2,0.5762432098388672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,8,power_law_1.01,0.249235200881958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,8,power_law_1.01,2.64333438873291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,8,power_law_1.01,1.988492774963379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,8,power_law_1.2,0.8008768081665039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,8,1,balanced,0.04376000165939331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,8,1,balanced,0.05806933343410492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,8,power_law_1.2,3.13885440826416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,8,1,balanced,0.10316800077756245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,8,1,balanced,0.1644373337427775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,8,1,balanced,0.16471999883651733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,8,1,balanced,0.1991306742032369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,8,1,balanced,0.16488533218701681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,8,1,balanced,0.20059732596079508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,8,1,balanced,0.1648426651954651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,8,1,balanced,0.20615466435750326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,8,1,balanced,0.1644533375898997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,8,power_law_1.01,2.8879104614257813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,8,1,balanced,0.25171200434366864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,8,1,balanced,0.16800532738367716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,8,1,balanced,0.25332266092300415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,8,1,balanced,0.2637066642443339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,8,1,balanced,0.28099199136098224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,8,1,balanced,0.29707199335098267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,8,1,balanced,0.32836800813674927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,8,power_law_1.01,0.324620795249939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,8,1,balanced,0.3628053267796834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,8,1,balanced,0.485152006149292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,8,power_law_1.01,0.47845120429992677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,8,1,balanced,0.5626879930496216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,8,power_law_1.2,1.004678440093994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,8,power_law_1.01,3.9053375244140627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,8,1,balanced,0.7814559936523438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,8,power_law_1.01,0.6915520191192627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,8,1,balanced,0.9939626852671305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,8,1,power_law_1.01,0.023263999819755556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,8,power_law_1.2,1.4250816345214843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,8,1,balanced,0.16973867019017538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,8,1,balanced,0.17108800013860068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,8,1,balanced,1.4271519978841145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,8,1,balanced,0.17708800236384073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,8,1,balanced,0.20348266760508218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,8,1,balanced,0.2057759960492452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,8,1,balanced,0.21758933862050375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,8,power_law_1.2,0.0828544020652771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,8,1,balanced,1.854101339975993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,8,power_law_1.2,1.8891839981079102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,8,power_law_1.2,0.12749439477920532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,8,1,power_law_1.01,0.0274944007396698
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,8,power_law_1.2,0.12295039892196655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,8,1,power_law_1.01,0.02967039942741394
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,8,power_law_1.2,0.11831680536270142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,8,1,power_law_1.01,0.04392960071563721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,8,power_law_1.2,4.02210578918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,8,1,power_law_1.01,0.05818880200386047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,8,1,power_law_1.01,0.02744320034980774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,8,1,power_law_1.01,0.0814848005771637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,8,1,power_law_1.01,0.047014400362968445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,8,1,power_law_1.01,0.10854400396347046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,8,1,balanced,0.2360960046450297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,8,1,power_law_1.01,0.06302719712257385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,8,1,balanced,0.25512532393137616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,8,1,balanced,0.29547733068466187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,8,1,power_law_1.01,0.09000959992408752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,8,1,balanced,0.33292800188064575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,8,1,balanced,0.467685341835022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,8,1,power_law_1.01,0.11461759805679321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,8,1,balanced,0.549295981725057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,8,1,power_law_1.01,0.15745919942855835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,8,1,balanced,0.766096035639445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,8,power_law_1.2,0.12985600233078004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,8,1,power_law_1.01,0.16783360242843628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,8,1,balanced,0.9845066865285238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,8,power_law_1.2,0.16881279945373534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,8,1,power_law_1.01,0.16880639791488647
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,8,power_law_1.2,0.2046207904815674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,8,1,balanced,1.4264480272928874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,8,1,power_law_1.01,0.1763200044631958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,8,power_law_1.2,0.2731136083602905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,8,1,power_law_1.01,0.18049919605255127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,8,1,power_law_1.01,0.12879999876022338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,8,1,balanced,1.8621919949849446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,8,power_law_1.2,0.3630143880844116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,8,1,power_law_1.01,0.13159040212631226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,8,1,power_law_1.2,0.023129600286483764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,8,power_law_1.2,0.48144001960754396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,8,1,power_law_1.01,0.13433599472045898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,8,1,power_law_1.2,0.030099201202392577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,8,power_law_1.01,3.3807552337646483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,8,1,power_law_1.01,0.13813120126724243
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,8,1,balanced,0.02735466758410136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,8,1,power_law_1.2,0.04684160053730011
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,8,1,balanced,0.027215999861558277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,8,1,power_law_1.2,0.025523200631141663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,8,1,power_law_1.01,0.14383360147476196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,8,1,power_law_1.2,0.06164479851722717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,8,1,power_law_1.2,0.03107840120792389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,8,1,power_law_1.01,0.15436160564422607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,8,1,power_law_1.2,0.08410879969596863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,8,1,power_law_1.2,0.04478079974651337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,8,1,power_law_1.01,0.1649791955947876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,8,1,power_law_1.01,0.02539519965648651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,8,1,power_law_1.2,0.10458240509033204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,8,1,power_law_1.2,0.05845119953155518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,8,1,power_law_1.01,0.1832576036453247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,8,1,power_law_1.01,0.18360320329666138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,8,1,power_law_1.01,0.025772801041603087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,8,1,power_law_1.2,0.16046080589294434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,8,1,power_law_1.01,0.19767040014266968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,8,1,power_law_1.01,0.18970240354537965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,8,1,power_law_1.01,0.028307199478149414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,8,1,power_law_1.2,0.1652799963951111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,8,1,power_law_1.01,0.20027520656585693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,8,power_law_1.2,0.798201608657837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,8,1,power_law_1.2,0.16558079719543456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,8,1,balanced,0.029552000264326733
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,8,1,balanced,0.03495999922355016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,8,1,balanced,0.05519466598828634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,8,1,balanced,0.08225599924723308
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,8,1,balanced,0.0805920014778773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,8,1,balanced,0.08247466882069905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,8,1,balanced,0.08387200037638347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,8,1,balanced,0.0855519970258077
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,8,1,balanced,0.08849066495895386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,8,1,balanced,0.09100799759229024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,8,1,balanced,0.09290666381518047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,8,1,balanced,0.09705066680908203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,8,1,power_law_1.2,0.07455999851226806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,8,1,power_law_1.01,0.2026304006576538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,8,1,power_law_1.2,0.09355520009994507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,8,power_law_1.2,2.696665573120117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,8,1,power_law_1.01,0.03653759956359863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,8,1,power_law_1.2,0.1250432014465332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,8,1,power_law_1.2,0.1763584017753601
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,8,1,power_law_1.01,0.044563201069831845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,8,1,power_law_1.2,0.13134080171585083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,8,1,power_law_1.2,0.17927039861679078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,8,1,power_law_1.01,0.06040319800376892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,8,1,power_law_1.2,0.13473279476165773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,8,1,power_law_1.2,0.18128639459609985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,8,1,power_law_1.01,0.06888960003852844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,8,1,power_law_1.2,0.13464319705963135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,8,1,power_law_1.2,0.18838399648666382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,8,1,power_law_1.2,0.14589439630508422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,8,power_law_1.2,3.639923095703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,8,1,balanced,0.10914132992426555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,8,1,power_law_1.2,0.20799360275268555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,8,1,balanced,0.11343466242154439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,8,1,power_law_1.2,0.15551359653472902
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,8,1,balanced,0.12496532996495564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,8,1,power_law_1.2,0.2416640043258667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,8,1,power_law_1.2,0.169977605342865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,8,1,power_law_1.01,0.21517438888549806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,8,1,power_law_1.2,0.18437119722366332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,8,1,power_law_1.01,0.23992960453033446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,8,1,power_law_1.2,0.19967999458312988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,8,1,power_law_1.2,0.024217599630355836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,8,1,power_law_1.2,0.20278398990631102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,8,1,power_law_1.2,0.21655681133270263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,8,1,power_law_1.01,0.07210239768028259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,8,1,power_law_1.2,0.24574079513549804
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,8,1,power_law_1.01,0.07552639842033386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,8,1,balanced,0.1267519990603129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,8,1,balanced,0.13993066549301147
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,8,1,power_law_1.01,0.07691519856452941
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,8,1,power_law_1.01,0.08133760094642639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,8,1,power_law_1.01,0.26746881008148193
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,8,1,power_law_1.2,0.024582399427890776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,8,1,power_law_1.01,0.323526406288147
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,8,1,power_law_1.2,0.027564799785614012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,8,1,power_law_1.01,0.37655680179595946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,8,1,power_law_1.2,0.27716479301452634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,8,1,power_law_1.2,0.03480960130691528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,8,1,power_law_1.01,0.4868159770965576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,8,1,power_law_1.2,0.24456319808959961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,8,1,power_law_1.2,0.3275840044021606
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,8,1,power_law_1.2,0.04168959856033325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,8,balanced,0.035589332381884255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,8,balanced,0.037690666814645134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,8,1,power_law_1.2,0.2571392059326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,8,1,power_law_1.01,0.5919487953186036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,8,1,power_law_1.2,0.3815551996231079
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,8,1,power_law_1.2,0.05857920050621033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,8,1,power_law_1.2,0.27572479248046877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,8,1,power_law_1.2,0.49358081817626953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,8,1,power_law_1.2,0.06968320012092591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,8,1,power_law_1.01,0.814463996887207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,8,1,power_law_1.2,0.30920960903167727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,8,1,power_law_1.01,0.08314239978790283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,8,1,power_law_1.2,0.5972671985626221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,8,1,power_law_1.2,0.365120005607605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,8,1,power_law_1.01,1.0322303771972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,8,1,power_law_1.2,0.40903677940368655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,8,1,power_law_1.2,0.8180031776428223
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,8,1,balanced,0.14999467134475708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,8,1,power_law_1.01,1.4750911712646484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,8,1,balanced,0.1686720053354899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,8,1,power_law_1.2,1.040287971496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,8,1,balanced,0.24215465784072876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,8,balanced,0.046181331078211464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,8,balanced,0.07047466437021892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,8,1,balanced,0.25093332926432294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,8,balanced,0.09039466579755147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,8,balanced,0.1239520013332367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,8,1,balanced,0.4193013509114583
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,8,1,balanced,0.37118931611378986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,8,1,power_law_1.2,1.4817407608032227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,8,1,power_law_1.01,1.9209728240966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,8,1,balanced,0.5179680188496908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,8,1,balanced,0.6687040328979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,8,power_law_1.01,0.06117119789123535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,8,1,power_law_1.01,0.2412800073623657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,8,power_law_1.01,0.08469120264053345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,8,power_law_1.01,0.08097280263900757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,8,1,power_law_1.01,0.24481918811798095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,8,1,power_law_1.2,1.9213056564331055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,8,power_law_1.01,0.08344320058822632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,8,1,power_law_1.01,0.25627520084381106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,8,power_law_1.01,0.05546240210533142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,8,balanced,0.12622933586438498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,8,power_law_1.01,0.1087488055229187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,8,balanced,0.12960533301035562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,8,1,power_law_1.01,0.2755327939987183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,8,power_law_1.01,0.07426559925079346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,8,power_law_1.01,0.115174400806427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,8,balanced,0.03558400024970373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,8,power_law_1.01,0.0641152024269104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,8,1,power_law_1.01,0.3039743900299072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,8,balanced,0.037690666814645134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,8,balanced,0.0397119993964831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,8,balanced,0.061679999033610024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,8,1,power_law_1.01,0.0870848000049591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,8,balanced,0.07723733286062877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,8,1,power_law_1.01,0.3615744113922119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,8,balanced,0.10962667067845662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,8,balanced,0.10970667004585266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,8,1,power_law_1.01,0.09410560131072998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,8,balanced,0.11409067114194234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,8,1,power_law_1.01,0.4038527965545654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,8,1,power_law_1.01,0.10056960582733154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,8,1,power_law_1.01,0.5027904033660888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,8,1,power_law_1.01,0.10801279544830322
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,8,1,power_law_1.01,0.12440320253372192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,8,1,power_law_1.01,0.6039936065673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,8,1,power_law_1.01,0.12174079418182374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,8,balanced,0.1280639966328939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,8,balanced,0.12852799892425537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,8,1,power_law_1.01,0.8203583717346191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,8,balanced,0.13343466321627298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,8,power_law_1.01,0.11995520591735839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,8,power_law_1.01,0.07200000286102295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,8,1,power_law_1.01,1.0336640357971192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,8,power_law_1.01,0.08469759821891784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,8,balanced,0.1156160036722819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,8,power_law_1.01,0.1050112009048462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,8,balanced,0.11450666189193726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,8,balanced,0.12065066893895467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,8,power_law_1.01,0.10528639554977418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,8,1,power_law_1.01,1.4728256225585938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,8,power_law_1.01,0.10875519514083862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,8,power_law_1.01,0.11112960577011108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,8,power_law_1.01,0.11306240558624267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,8,1,power_law_1.01,0.13950719833374023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,8,1,power_law_1.2,0.07171199917793274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,8,1,power_law_1.01,1.9051200866699218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,8,balanced,0.13452800114949545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,8,balanced,0.13784533739089966
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,8,1,power_law_1.01,0.1449280023574829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,8,power_law_1.01,0.12035200595855713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,8,balanced,0.14586666226387024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,8,balanced,0.14992533127466837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,8,1,power_law_1.01,0.17077120542526245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,8,balanced,0.15590932965278625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,8,balanced,0.1686506668726603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,8,balanced,0.20181334018707275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,8,1,power_law_1.2,0.5176127910614013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,8,balanced,0.2344800035158793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,8,balanced,0.12377599875132243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,8,balanced,0.3699359893798828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,8,balanced,0.1285760005315145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,8,balanced,0.13783466815948486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,8,balanced,0.14434132973353067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,8,balanced,0.15482667088508606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,8,balanced,0.17123200496037802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,8,balanced,0.20568533738454184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,8,balanced,0.2381653388341268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,8,power_law_1.01,0.11737600564956666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,8,balanced,0.3533173402150472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,8,power_law_1.01,0.12037119865417481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,8,power_law_1.2,0.06255360245704651
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,8,1,power_law_1.2,0.07504000067710877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,8,power_law_1.01,0.12391680479049683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,8,power_law_1.01,0.12082560062408447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,8,1,power_law_1.2,0.07551360130310059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,8,power_law_1.01,0.1362239956855774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,8,power_law_1.01,0.12224639654159546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,8,1,power_law_1.2,0.0802944004535675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,8,power_law_1.01,0.149452805519104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,8,power_law_1.01,0.1252351999282837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,8,1,power_law_1.2,0.08286719918251037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,8,power_law_1.01,0.13000960350036622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,8,balanced,0.4320106506347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,8,1,power_law_1.2,0.08768640160560608
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,8,1,power_law_1.2,0.09301120042800903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,8,1,power_law_1.2,0.10046720504760742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,8,balanced,0.42028268178304035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,8,1,power_law_1.2,0.10736000537872314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,8,balanced,0.5788640181223551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,8,power_law_1.2,0.07667199969291687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,8,1,power_law_1.2,0.12537599802017213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,8,balanced,0.7170453071594238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,8,1,power_law_1.01,0.22021760940551757
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,8,1,power_law_1.2,0.12345600128173828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,8,balanced,1.0483946800231934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,8,1,power_law_1.2,0.1403839945793152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,8,1,power_law_1.2,0.6112127780914307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,8,balanced,1.3507466316223145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,8,1,power_law_1.2,0.830726432800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,8,balanced,1.9811360041300456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,8,1,power_law_1.2,1.0465984344482422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,8,balanced,2.6140054066975913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,8,power_law_1.2,0.07935360074043274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,8,1,power_law_1.2,1.4849727630615235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,8,power_law_1.2,0.0856383979320526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,8,1,power_law_1.01,0.26936318874359133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,8,power_law_1.01,0.16603519916534423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,8,power_law_1.2,0.08926720023155213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,8,1,power_law_1.01,0.3208127975463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,8,power_law_1.01,0.1852031946182251
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,8,1,power_law_1.2,0.14602240324020385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,8,power_law_1.2,0.11724159717559815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,8,power_law_1.01,0.23710079193115235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,8,1,power_law_1.01,0.39981439113616946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,8,power_law_1.2,0.11635839939117432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,8,1,power_law_1.2,1.926291275024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,8,1,balanced,0.025392000873883564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,8,power_law_1.2,0.12020479440689087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,8,1,balanced,0.03421333432197571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,8,1,balanced,0.05179733534653982
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,8,1,power_law_1.01,0.5477375984191895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,8,power_law_1.2,0.11957759857177734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,8,power_law_1.2,0.12311040163040161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,8,power_law_1.2,0.12825599908828736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,8,1,power_law_1.01,0.7747968196868896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,8,1,balanced,0.03158933420976003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,8,power_law_1.2,0.12577919960021972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,8,1,balanced,0.03452266752719879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,8,power_law_1.2,0.054451197385787964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,8,power_law_1.2,0.13050880432128906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,8,power_law_1.2,0.06454399824142457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,8,power_law_1.01,0.13222399950027466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,8,power_law_1.2,0.07052800059318542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,8,power_law_1.01,0.13974399566650392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,8,1,power_law_1.2,0.17356159687042236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,8,power_law_1.2,0.08358399868011475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,8,power_law_1.01,0.1664512038230896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,8,power_law_1.01,0.2803839921951294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,8,power_law_1.2,0.08057600259780884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,8,power_law_1.01,0.18010239601135253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,8,1,balanced,0.09132267038027446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,8,power_law_1.01,0.3746239900588989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,8,1,balanced,0.14853866895039877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,8,power_law_1.01,0.20069119930267335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,8,1,balanced,0.20313066244125366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,8,1,balanced,0.20438933372497559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,8,power_law_1.01,0.46015357971191406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,8,power_law_1.01,0.24902400970458985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,8,1,balanced,0.04614933331807455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,8,power_law_1.01,0.6404607772827149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,8,1,balanced,0.084906667470932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,8,1,balanced,0.1429333289464315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,8,1,balanced,0.1856693426767985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,8,balanced,0.5971146821975708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,8,1,balanced,0.18673600753148398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,8,power_law_1.01,0.8222463607788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,8,1,balanced,0.18773865699768066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,8,balanced,0.7275573412577311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,8,balanced,1.06660262743632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,8,power_law_1.01,1.2375295639038086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,8,power_law_1.2,0.10554879903793335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,8,balanced,1.3632853825887044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,8,power_law_1.2,0.10542080402374268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,8,1,balanced,0.20354666312535605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,8,power_law_1.01,1.4877311706542968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,8,power_law_1.01,0.2987839937210083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,8,balanced,2.0373600323994956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,8,power_law_1.01,0.40417280197143557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,8,power_law_1.2,0.14041600227355958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,8,power_law_1.2,0.17556480169296265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,8,power_law_1.01,0.49511041641235354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,8,power_law_1.01,2.3903680801391602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,8,balanced,2.80513064066569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,8,power_law_1.2,0.1865664005279541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,8,1,power_law_1.2,0.22167680263519288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,8,power_law_1.01,0.6703936100006104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,8,power_law_1.2,0.20690560340881348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,8,1,power_law_1.01,0.02436479926109314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,8,1,power_law_1.2,0.27752320766448973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,8,power_law_1.2,0.2565824031829834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,8,1,power_law_1.01,0.034790399670600894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,8,power_law_1.01,0.8544704437255859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,8,1,power_law_1.01,0.054054397344589236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,8,power_law_1.2,0.10917119979858399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,8,power_law_1.01,3.015488052368164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,8,1,power_law_1.01,0.08082559704780579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,8,power_law_1.2,0.1106495976448059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,8,power_law_1.01,1.256608009338379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,8,1,power_law_1.01,0.028198400139808656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,8,1,power_law_1.01,0.0350271999835968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,8,1,balanced,0.18894400199254355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,8,1,power_law_1.01,0.05249919891357422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,8,power_law_1.01,1.6192512512207031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,8,power_law_1.2,0.32844159603118894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,8,1,balanced,0.20484799146652222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,8,1,balanced,0.20484799146652222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,8,power_law_1.01,2.362681579589844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,8,power_law_1.2,0.41535358428955077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,8,power_law_1.2,0.11299840211868287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,8,power_law_1.2,0.5139711856842041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,8,power_law_1.2,0.1159551978111267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,8,power_law_1.2,0.12174079418182374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,8,1,balanced,0.18765334288279215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,8,power_law_1.2,0.7279808044433593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,8,1,balanced,0.1901386578877767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,8,power_law_1.2,0.12561919689178466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,8,1,power_law_1.01,0.07795199751853943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,8,power_law_1.01,3.1786752700805665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,8,power_law_1.2,0.14611200094223023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,8,1,power_law_1.01,0.09968640208244324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,8,power_law_1.2,0.15299839973449708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,8,1,power_law_1.2,0.328659200668335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,8,power_law_1.2,0.17155840396881103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,8,1,balanced,0.20516266425450644
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,8,1,power_law_1.2,0.40607361793518065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,8,1,balanced,0.23004267613093057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,8,1,power_law_1.01,0.10456960201263428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,8,1,balanced,0.23144533236821493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,8,1,balanced,0.23670933643976846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,8,1,power_law_1.01,0.15676159858703614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,8,1,power_law_1.2,0.5521728038787842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,8,1,balanced,0.23668267329533896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,8,1,balanced,0.24187199274698892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,8,1,balanced,0.24652800957361856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,8,1,balanced,0.26525332530339557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,8,1,balanced,0.2016693353652954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,8,1,power_law_1.2,0.8179648399353028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,8,1,balanced,0.274399995803833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,8,1,balanced,0.19971734285354614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,8,1,balanced,0.20562134186426798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,8,1,power_law_1.2,0.028774398565292358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,8,1,power_law_1.2,0.02428800016641617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,8,1,power_law_1.2,0.036620798707008365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,8,power_law_1.2,0.9532544136047363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,8,1,power_law_1.2,0.04147199988365173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,8,1,power_law_1.2,0.05189759731292724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,8,1,power_law_1.2,0.05262079834938049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,8,power_law_1.2,0.19153280258178712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,8,1,power_law_1.2,0.07212160229682922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,8,1,power_law_1.2,0.07650560140609741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,8,power_law_1.2,0.23766400814056396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,8,1,power_law_1.2,0.0942080020904541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,8,1,power_law_1.2,0.09366400241851806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,8,power_law_1.2,0.2955712080001831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,8,1,power_law_1.2,0.1419711947441101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,8,1,power_law_1.01,0.15928319692611695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,8,1,power_law_1.2,0.15511679649353027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,8,1,power_law_1.01,0.16701439619064332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,8,1,power_law_1.2,0.16457600593566896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,8,1,power_law_1.01,0.17008639574050904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,8,1,power_law_1.2,0.17121920585632325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,8,1,balanced,0.2109973430633545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,8,1,power_law_1.01,0.14520959854125975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,8,1,balanced,0.21329599618911743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,8,1,power_law_1.2,0.18072960376739503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,8,1,balanced,0.22683199246724448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,8,1,power_law_1.01,0.16111359596252442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,8,1,power_law_1.2,0.1916800022125244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,8,1,power_law_1.01,0.163372802734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,8,1,power_law_1.2,0.18825600147247315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,8,power_law_1.2,1.3547840118408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,8,1,power_law_1.2,0.19623039960861205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,8,1,power_law_1.2,0.150764799118042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,8,power_law_1.2,0.39187839031219485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,8,1,power_law_1.2,0.15991679430007935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,8,1,power_law_1.2,0.16583679914474486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,8,1,balanced,0.44727468490600586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,8,1,power_law_1.2,0.18081920146942138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,8,1,balanced,0.4645013411839803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,8,1,power_law_1.2,0.18948479890823364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,8,1,balanced,0.23666133483250937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,8,1,balanced,0.259333332379659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,8,1,balanced,0.40996265411376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,8,1,power_law_1.01,0.16695040464401245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,8,1,power_law_1.01,0.17329280376434325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,8,1,power_law_1.2,0.20802559852600097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,8,1,power_law_1.2,0.2206592082977295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,8,1,power_law_1.2,0.22450559139251708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,8,1,power_law_1.01,0.187116801738739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,8,1,power_law_1.2,0.25283839702606203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,8,1,balanced,0.6443093220392863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,8,1,power_law_1.01,0.203603196144104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,8,1,power_law_1.2,0.28136320114135743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,8,1,power_law_1.2,0.20443520545959473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,8,1,power_law_1.01,0.21298561096191407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,8,1,power_law_1.2,0.3148159980773926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,8,1,power_law_1.2,0.21492478847503663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,8,1,power_law_1.01,0.21575040817260743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,8,1,power_law_1.2,0.3919296026229858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,8,1,power_law_1.2,0.2154560089111328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,8,1,power_law_1.2,0.46160001754760743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,8,1,power_law_1.2,0.23515520095825196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,8,power_law_1.2,1.7150848388671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,8,1,power_law_1.2,0.25177600383758547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,8,1,power_law_1.2,0.6082111835479737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,8,power_law_1.2,0.4826943874359131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,8,1,power_law_1.2,0.2778879880905151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,8,1,power_law_1.2,0.27720320224761963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,8,power_law_1.2,0.6729087829589844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,8,1,power_law_1.2,0.3184959888458252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,8,1,balanced,0.6995360056559244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,8,power_law_1.2,0.8632063865661621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,8,1,power_law_1.2,0.3513216018676758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,8,1,balanced,0.43662933508555096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,8,1,balanced,0.613040010134379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,8,1,power_law_1.01,0.22257919311523439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,8,1,power_law_1.01,0.18976000547409058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,8,1,power_law_1.01,0.23681919574737548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,8,1,power_law_1.01,0.18721280097961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,8,1,power_law_1.01,0.24316160678863524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,8,1,power_law_1.01,0.18985600471496583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,8,1,power_law_1.01,0.283353590965271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,8,1,power_law_1.01,0.20215039253234862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,8,1,power_law_1.01,0.31146879196166993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,8,1,power_law_1.01,0.2120959997177124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,8,1,power_law_1.01,0.3457024097442627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,8,power_law_1.2,2.6540096282958983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,8,1,power_law_1.01,0.21775999069213867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,8,1,power_law_1.2,0.7510015964508057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,8,1,power_law_1.01,0.41656317710876467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,8,1,balanced,1.063167969385783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,8,1,power_law_1.2,0.4214975833892822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,8,1,power_law_1.01,0.49498882293701174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,8,1,balanced,1.303968032201131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,8,1,balanced,0.677226702372233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,8,1,power_law_1.2,1.0489791870117187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,8,1,power_law_1.2,0.4992640018463135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,8,1,power_law_1.01,0.6321792125701904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,8,1,balanced,1.0268959999084473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,8,1,balanced,1.884592056274414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,8,1,power_law_1.2,0.6362815856933594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,8,1,power_law_1.2,1.3405247688293458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,8,1,power_law_1.01,0.7759359836578369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,8,1,balanced,1.2686560153961182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,8,1,power_law_1.2,0.7867648124694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,8,1,balanced,2.482149283091227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,8,1,power_law_1.01,1.0728320121765136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,8,balanced,0.08063466846942902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,8,balanced,0.11909332871437073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,8,1,power_law_1.01,0.2427135944366455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,8,balanced,0.1901866594950358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,8,balanced,0.3296533425649007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,8,1,power_law_1.01,0.2771712064743042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,8,1,power_law_1.01,1.3706879615783691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,8,balanced,0.5637653271357218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,8,1,power_law_1.01,0.3067008018493652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,8,balanced,0.7421387036641439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,8,1,power_law_1.01,0.38156800270080565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,8,balanced,0.7499306996663412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,8,balanced,0.7574346860249838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,8,1,power_law_1.01,0.45525760650634767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,8,balanced,1.0344266891479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,8,1,power_law_1.01,0.601203203201294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,8,1,balanced,1.8533387184143066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,8,balanced,1.0439146359761555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,8,power_law_1.2,1.3065664291381835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,8,1,power_law_1.2,1.0787648200988769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,8,1,power_law_1.01,0.7438271999359131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,8,balanced,1.0645066897074382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,8,1,balanced,2.4457972844441733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,8,balanced,1.0846880276997883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,8,1,power_law_1.01,1.0425344467163087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,8,power_law_1.2,1.6673343658447266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,8,balanced,0.07027733325958252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,8,balanced,0.10145599643389384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,8,1,power_law_1.2,1.3694399833679198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,8,balanced,0.1458613375822703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,8,balanced,0.23449599742889404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,8,balanced,0.3908533255259196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,8,1,power_law_1.01,1.3329279899597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,8,balanced,0.523141344388326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,8,balanced,0.5334293444951376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,8,balanced,0.5438559850056967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,8,1,power_law_1.01,1.9199039459228515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,8,balanced,0.8357973098754883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,8,balanced,0.8465279738108317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,8,balanced,0.8694667021433512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,8,balanced,0.8921333154042562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,8,1,power_law_1.01,2.512620735168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,8,balanced,0.9182986418406168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,8,1,power_law_1.01,1.951590347290039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,8,balanced,0.956501324971517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,8,balanced,0.9964053630828857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,8,balanced,0.0793333351612091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,8,balanced,0.10101333260536194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,8,1,power_law_1.01,2.5431615829467775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,8,balanced,0.14010666807492575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,8,balanced,0.25733333826065063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,8,balanced,0.44949865341186523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,8,power_law_1.01,0.2923968076705933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,8,balanced,0.6159199873606364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,8,balanced,1.106767972310384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,8,balanced,0.627888003985087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,8,balanced,1.143338680267334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,8,balanced,0.63645867506663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,8,power_law_1.01,0.5130047798156738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,8,balanced,0.6473226547241211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,8,balanced,1.1815733114878337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,8,balanced,0.6600586573282877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,8,balanced,1.0445066293080647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,8,balanced,0.6784533659617106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,8,1,power_law_1.2,1.9603647232055663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,8,balanced,1.1292479832967122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,8,1,power_law_1.2,1.9338560104370117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,8,balanced,1.2979146639506023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,8,balanced,1.462117354075114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,8,1,power_law_1.2,2.5270463943481447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,8,balanced,2.1100266774495444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,8,power_law_1.01,0.4415616035461426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,8,balanced,1.2236693700154622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,8,power_law_1.01,0.19328000545501708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,8,balanced,0.6991253693898519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,8,balanced,1.2985813617706299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,8,power_law_1.01,0.5554431915283203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,8,balanced,0.7224693298339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,8,balanced,2.4340160687764487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,8,power_law_1.01,0.3386431932449341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,8,balanced,1.4464267094930012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,8,balanced,0.7609066963195801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,8,power_law_1.01,0.6665535926818847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,8,power_law_1.01,0.3076416015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,8,balanced,1.6018667221069336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,8,balanced,3.4307680130004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,8,power_law_1.01,0.663974380493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,8,power_law_1.01,0.38867199420928955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,8,balanced,2.20197327931722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,8,power_law_1.01,0.6191552162170411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,8,power_law_1.01,0.4522751808166504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,8,balanced,2.5181280771891275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,8,balanced,4.1445919672648115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,8,power_law_1.01,0.48159360885620117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,8,power_law_1.01,0.7106048107147217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,8,balanced,0.7998720010121664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,8,power_law_1.01,0.49441280364990237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,8,balanced,0.8394560019175211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,8,power_law_1.01,0.9461440086364746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,8,balanced,3.719562530517578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,8,power_law_1.01,0.49303040504455564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,8,power_law_1.01,0.9421440124511719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,8,power_law_1.01,0.7790592193603516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,8,balanced,6.540128072102864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,8,power_law_1.01,1.0007807731628418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,8,power_law_1.01,0.7877376079559326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,8,balanced,4.6397705078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,8,power_law_1.01,1.0515263557434082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,8,power_law_1.2,3.4068416595458983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,8,power_law_1.01,0.8378047943115234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,8,power_law_1.01,1.0834752082824708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,8,power_law_1.01,0.8529664039611816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,8,power_law_1.01,0.8869119644165039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,8,power_law_1.01,0.1984063982963562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,8,balanced,7.027733484903972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,8,power_law_1.01,0.9386112213134765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,8,power_law_1.01,0.3278464078903198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,8,power_law_1.01,0.9940223693847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,8,power_law_1.01,0.31008639335632326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,8,balanced,0.9184799989064535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,8,power_law_1.01,1.0409152030944824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,8,balanced,1.064245303471883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,8,power_law_1.01,0.3805504083633423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,8,power_law_1.01,1.1341567993164063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,8,balanced,1.2155626614888508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,8,power_law_1.01,1.180992031097412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,8,power_law_1.01,0.4869376182556152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,8,power_law_1.01,1.1554304122924806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,8,balanced,1.7552746136983235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,8,balanced,9.142080307006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,8,power_law_1.01,1.4491711616516114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,8,power_law_1.01,0.6114687919616699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,8,power_law_1.01,1.2049280166625977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,8,balanced,2.05132261912028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,8,power_law_1.01,0.5859519958496093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,8,power_law_1.01,1.703321647644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,8,power_law_1.01,1.3474495887756348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,8,power_law_1.01,0.5987904071807861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,8,balanced,2.908405303955078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,8,balanced,8.660309473673502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,8,power_law_1.01,0.5917312145233155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,8,1,power_law_1.2,2.548588752746582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,8,power_law_1.01,2.245587158203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,8,power_law_1.01,0.6065855979919433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,8,balanced,3.5016959508260093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,8,power_law_1.01,0.6286784172058105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,8,power_law_1.01,2.6867136001586913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,8,power_law_1.01,0.6722112178802491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,8,power_law_1.2,2.5556032180786135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,8,power_law_1.01,0.6793920040130615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,8,power_law_1.01,1.5218815803527832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,8,balanced,5.223168055216472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,8,power_law_1.01,0.745363187789917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,8,power_law_1.01,3.6349952697753904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,8,power_law_1.2,0.291430401802063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,8,power_law_1.01,1.8022272109985351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,8,power_law_1.01,0.7874688148498535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,8,power_law_1.2,0.40245761871337893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,8,balanced,13.208175659179688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,8,power_law_1.01,0.8455552101135254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,8,power_law_1.2,0.44936318397521974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,8,balanced,13.714730580647787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,8,power_law_1.01,2.2204479217529296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,8,balanced,6.668229420979817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,8,power_law_1.2,3.361036682128906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,8,power_law_1.01,0.9561663627624511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,8,power_law_1.2,0.542905616760254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,8,power_law_1.01,4.693158340454102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,8,power_law_1.01,1.1630144119262695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,8,power_law_1.2,0.6481855869293213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,8,power_law_1.01,2.6989696502685545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,8,power_law_1.01,1.383187198638916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,8,power_law_1.2,0.7366911888122558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,8,power_law_1.2,0.6212031841278076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,8,power_law_1.01,1.835001564025879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,8,power_law_1.01,3.664332962036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,8,balanced,9.868848164876303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,8,power_law_1.01,6.510771179199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,8,power_law_1.2,0.19456640481948853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,8,power_law_1.2,0.24849278926849366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,8,power_law_1.01,2.2335615158081055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,8,power_law_1.2,0.29326720237731935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,8,balanced,17.249322255452473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,8,power_law_1.2,0.3487231969833374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,8,power_law_1.01,4.727328109741211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,8,balanced,17.96291224161784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,8,power_law_1.01,3.1128320693969727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,8,power_law_1.2,0.6895743846893311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,8,1,balanced,0.07693866888682048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,8,1,balanced,0.09382933378219604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,8,1,balanced,0.17085866133371988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,8,1,balanced,0.29803733030955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,8,power_law_1.2,1.0042559623718261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,8,1,balanced,0.5390239953994751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,8,power_law_1.01,8.384966278076172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,8,power_law_1.2,1.0135104179382324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,8,power_law_1.2,0.1983296036720276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,8,balanced,12.756095886230469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,8,power_law_1.01,3.9676097869873046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,8,power_law_1.2,1.0490816116333008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,8,power_law_1.2,0.24922239780426025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,8,power_law_1.01,6.486918640136719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,8,power_law_1.2,1.025222396850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,8,power_law_1.2,1.0865792274475097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,8,1,balanced,0.7729439735412598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,8,1,balanced,0.7759040196736654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,8,power_law_1.2,1.1374655723571778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,8,1,balanced,0.7796693642934164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,8,power_law_1.01,5.739968109130859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,8,1,balanced,0.06320533156394958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,8,1,balanced,0.08001066744327545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,8,1,balanced,0.1320319970448812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,8,1,balanced,0.22007467349370322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,8,power_law_1.2,0.4413248062133789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,8,power_law_1.2,0.4690112113952637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,8,power_law_1.01,8.582943725585938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,8,power_law_1.01,12.711808013916016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,8,power_law_1.2,0.48090238571166993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,8,1,balanced,0.7874879837036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,8,1,balanced,0.7879093488057455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,8,power_law_1.2,1.2036288261413575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,8,1,balanced,0.39267198244730633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,8,power_law_1.01,7.191474914550781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,8,1,balanced,0.5562453269958496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,8,1,balanced,0.5609920024871826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,8,power_law_1.2,1.2699328422546388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,8,1,balanced,0.5641493399937948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,8,1,balanced,0.5660266478856405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,8,1,balanced,0.5688800017038981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,8,power_law_1.2,1.381932830810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,8,1,balanced,0.575434684753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,8,power_law_1.2,0.4967360019683838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,8,1,balanced,0.5818826754887899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,8,1,balanced,0.5944960117340088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,8,power_law_1.2,0.7760640144348144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,8,1,balanced,0.6051466862360636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,8,1,balanced,0.7998560269673666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,8,power_law_1.2,1.6140863418579101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,8,1,balanced,0.6168853441874186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,8,1,balanced,0.8049973646799723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,8,power_law_1.2,0.8056575775146484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,8,1,balanced,0.8592373530069987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,8,1,balanced,0.8190666834513346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,8,1,balanced,0.8296746412913004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,8,1,balanced,1.1185493469238281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,8,power_law_1.2,1.800115203857422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,8,power_law_1.2,0.8402303695678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,8,1,balanced,0.8445226351420084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,8,1,balanced,0.8525280157725016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,8,power_law_1.01,13.789907836914063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,8,power_law_1.2,0.8552512168884278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,8,1,balanced,1.0862773259480794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,8,power_law_1.01,10.83514862060547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,8,power_law_1.2,2.4475839614868162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,8,power_law_1.2,0.8927359580993652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,8,power_law_1.2,0.28819200992584226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,8,power_law_1.2,0.3729599952697754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,8,power_law_1.01,17.277357482910155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,8,power_law_1.2,2.859744071960449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,8,power_law_1.2,0.4792640209197998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,8,1,power_law_1.01,0.0639743983745575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,8,1,balanced,0.9407306512196859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,8,1,power_law_1.01,0.09117439985275269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,8,1,balanced,1.3950506846110027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,8,1,balanced,1.4349652926127117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,8,1,balanced,0.9827893575032552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,8,1,power_law_1.01,0.16814719438552855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,8,power_law_1.2,3.7219966888427733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,8,1,balanced,1.5924372673034668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,8,1,power_law_1.01,0.25754239559173586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,8,1,balanced,2.2675466537475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,8,power_law_1.2,0.5880640029907227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,8,1,balanced,2.6126453081766763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,8,power_law_1.2,0.5504960060119629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,8,power_law_1.2,4.850630569458008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,8,power_law_1.2,0.966214370727539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,8,power_law_1.2,0.6076543807983399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,8,power_law_1.01,14.996383666992188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,8,1,balanced,3.9279680252075195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,8,power_law_1.2,0.607206392288208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,8,1,power_law_1.01,0.372761607170105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,8,power_law_1.01,18.002029418945312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,8,1,power_law_1.01,0.05514240264892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,8,power_law_1.2,0.6097023963928223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,8,1,power_law_1.01,0.5978752136230469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,8,1,power_law_1.2,0.06499199867248535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,8,1,power_law_1.01,0.07575039863586426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,8,power_law_1.2,0.6294847965240479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,8,1,power_law_1.01,0.618227195739746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,8,1,power_law_1.2,0.09276800155639649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,8,1,balanced,4.932544072469075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,8,1,power_law_1.01,0.12815359830856324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,8,power_law_1.2,0.6660863876342773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,8,1,power_law_1.2,0.16752640008926392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,8,1,power_law_1.01,0.641977596282959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,8,1,power_law_1.01,0.20364160537719728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,8,power_law_1.2,1.0346559524536132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,8,1,power_law_1.2,0.2474047899246216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,8,1,power_law_1.01,0.276691198348999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,8,1,power_law_1.2,0.35280001163482666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,8,power_law_1.2,1.0951616287231445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,8,1,power_law_1.01,0.43405442237854003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,8,1,balanced,7.28219731648763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,8,1,power_law_1.2,0.5846144199371338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,8,1,power_law_1.01,0.4574399948120117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,8,power_law_1.2,1.2053567886352539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,8,power_law_1.2,6.982649230957032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,8,1,power_law_1.01,0.47827839851379395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,8,power_law_1.2,0.7008255958557129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,8,power_law_1.2,1.4529343605041505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,8,power_law_1.2,0.7537792205810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,8,power_law_1.2,0.8040063858032227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,8,1,balanced,1.0435573259989421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,8,power_law_1.2,0.8404031753540039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,8,1,balanced,9.331504185994467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,8,1,power_law_1.2,0.6136703968048096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,8,1,balanced,1.5163413683573406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,8,power_law_1.2,0.9638208389282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,8,1,power_law_1.01,0.49152002334594724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,8,1,power_law_1.2,0.6414400100708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,8,1,balanced,1.6522560119628906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,8,1,power_law_1.01,0.5053120136260987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,8,power_law_1.2,1.1852607727050781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,8,1,power_law_1.2,0.6838592052459717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,8,power_law_1.2,1.7171007156372071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,8,1,power_law_1.01,0.556108808517456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,8,1,balanced,2.3062240282694497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,8,1,power_law_1.2,0.7007552146911621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,8,1,power_law_1.01,0.5920703887939454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,8,power_law_1.2,1.4149888038635254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,8,1,power_law_1.2,0.7439487934112549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,8,1,balanced,2.6247733434041343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,8,1,power_law_1.01,0.6466559886932373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,8,power_law_1.2,2.244108772277832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,8,1,power_law_1.01,0.6771967887878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,8,1,power_law_1.2,0.052742397785186766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,8,1,power_law_1.01,0.6860544204711914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,8,1,power_law_1.2,0.07646719813346863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,8,1,balanced,3.9276533126831055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,8,power_law_1.2,2.76376953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,8,power_law_1.2,9.929529571533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,8,1,power_law_1.01,0.7195839881896973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,8,1,power_law_1.01,0.7666687965393066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,8,1,power_law_1.01,0.8204928398132324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,8,1,power_law_1.2,0.7909503936767578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,8,1,balanced,5.002485275268555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,8,1,power_law_1.01,0.7314239978790283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,8,power_law_1.2,3.83741455078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,8,1,power_law_1.01,0.9219584465026855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,8,1,power_law_1.2,0.8514752388000488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,8,1,power_law_1.01,0.8244288444519043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,8,1,power_law_1.2,0.12681599855422973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,8,1,power_law_1.01,1.0318016052246093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,8,1,power_law_1.01,0.922111988067627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,8,1,power_law_1.2,0.9532032012939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,8,1,power_law_1.2,0.18485759496688842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,8,1,power_law_1.01,1.143603229522705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,8,1,power_law_1.01,1.106867218017578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,8,1,power_law_1.2,1.0627072334289551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,8,1,power_law_1.2,0.2506688117980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,8,power_law_1.2,4.983110427856445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,8,power_law_1.2,1.8560640335083007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,8,1,power_law_1.01,0.9338111877441406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,8,1,power_law_1.2,1.180121612548828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,8,1,power_law_1.2,0.42239999771118164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,8,1,power_law_1.01,1.073747158050537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,8,1,power_law_1.2,0.45087361335754395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,8,power_law_1.2,2.302195167541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,8,1,power_law_1.2,1.4002047538757325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,8,1,power_law_1.2,0.47205119132995604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,8,1,power_law_1.01,1.3481151580810546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,8,1,power_law_1.2,1.051238441467285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,8,1,power_law_1.2,0.49984002113342285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,8,1,balanced,7.373114903767903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,8,1,power_law_1.01,1.3649279594421386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,8,1,power_law_1.2,0.5268159866333008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,8,1,power_law_1.2,1.1767616271972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,8,power_law_1.2,6.909951782226562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,8,1,power_law_1.2,0.573356819152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,8,1,power_law_1.01,1.0288895606994628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,8,1,power_law_1.2,1.4333439826965333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,8,1,power_law_1.2,0.6151999950408935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,8,1,power_law_1.01,1.1541824340820312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,8,1,power_law_1.2,0.6674752235412598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,8,1,power_law_1.2,1.686195182800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,8,1,power_law_1.01,1.4089280128479005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,8,power_law_1.2,16.38657989501953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,8,1,power_law_1.01,1.6259071350097656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,8,power_law_1.2,3.223775863647461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,8,1,balanced,9.336975733439127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,8,1,power_law_1.2,0.7674431800842285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,16,1,balanced,0.030063999195893604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,16,1,balanced,0.05182399849096934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,8,1,power_law_1.2,2.215385627746582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,16,1,balanced,0.062362665931383766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,16,1,balanced,0.06344533463319142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,8,1,power_law_1.2,0.8542783737182618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,8,1,power_law_1.01,2.176998329162598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,8,power_law_1.2,9.436653137207031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,8,1,power_law_1.2,0.9478079795837402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,8,1,power_law_1.2,2.765318489074707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,8,1,power_law_1.01,2.740620803833008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,8,1,power_law_1.2,1.131392002105713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,8,1,power_law_1.01,1.6780031204223633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,8,1,power_law_1.2,0.9490880012512207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,16,1,balanced,0.09286399682362874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,16,1,balanced,0.09321066737174988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,16,1,balanced,0.0941973328590393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,16,1,balanced,0.09341866771380107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,16,1,balanced,0.09582933783531189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,16,1,balanced,0.09703999757766724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,16,1,balanced,0.09784000118573506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,16,1,balanced,0.10123200217882793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,8,1,power_law_1.01,2.205779266357422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,16,1,balanced,0.10204799969991048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,8,power_law_1.2,4.090496063232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,8,1,power_law_1.01,2.7598527908325194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,8,1,power_law_1.2,3.843859100341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,8,1,power_law_1.2,1.1009535789489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,8,1,power_law_1.01,3.844812774658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,16,1,balanced,0.11061333616574605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,16,1,balanced,0.11742933591206868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,8,power_law_1.2,6.134534454345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,8,1,power_law_1.2,1.373420810699463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,8,1,power_law_1.01,3.8310848236083985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,8,1,power_law_1.2,1.6418367385864259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,8,1,power_law_1.2,4.9338432312011715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,8,1,power_law_1.01,4.966022491455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,8,1,power_law_1.2,2.2006591796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,16,1,balanced,0.1239359974861145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,8,power_law_1.2,13.498565673828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,16,1,balanced,0.11598400274912517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,16,1,balanced,0.13570666313171387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,16,1,balanced,0.16456533471743265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,16,1,balanced,0.23140267531077066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,16,1,balanced,0.2647999922434489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,8,power_law_1.2,7.677836608886719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,8,1,power_law_1.2,2.759929656982422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,8,1,power_law_1.2,7.521478271484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,8,1,power_law_1.01,7.2372291564941404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,8,1,power_law_1.2,3.8631744384765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,16,1,balanced,0.37149866422017414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,16,1,balanced,0.4755093256632487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,16,1,balanced,0.6898773511250814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,16,1,balanced,0.903439998626709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,16,1,balanced,1.3356159528096516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,8,1,power_law_1.01,5.057952117919922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,8,1,power_law_1.2,4.987507247924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,16,1,balanced,1.7659200032552083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,8,power_law_1.2,11.376697540283203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,8,1,power_law_1.01,9.484268951416016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,16,1,balanced,0.03979199876387914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,16,1,balanced,0.043562665581703186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,16,1,balanced,0.05206400156021118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,16,1,balanced,0.054042667150497437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,16,1,balanced,0.05406400064627329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,16,1,balanced,0.05539200206597646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,16,1,balanced,0.05179733534653982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,16,1,balanced,0.052042668064435325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,16,1,balanced,0.05555733541647593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,16,1,balanced,0.05349333087603251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,16,1,balanced,0.05821333328882853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,16,1,balanced,0.06196266909440359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,8,1,power_law_1.01,7.2668609619140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,16,1,balanced,0.06461866696675618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,16,1,balanced,0.07232533395290375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,16,1,balanced,0.08509332935015361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,8,1,power_law_1.2,7.2541053771972654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,16,1,balanced,0.08850666880607605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,16,1,balanced,0.08656000097592671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,16,1,balanced,0.12160000205039978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,16,1,balanced,0.13918933272361755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,16,1,balanced,0.192303995291392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,16,1,balanced,0.2384000023206075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,16,1,balanced,0.3261013428370158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,16,1,balanced,0.4219893217086792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,16,1,balanced,0.6042079925537109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,16,1,balanced,0.7901013692220052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,16,1,balanced,0.03345600018898646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,16,1,balanced,0.03754666695992152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,16,1,balanced,0.05468800167242686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,16,1,balanced,0.05395199855168661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,16,1,balanced,1.1542719999949138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,16,1,balanced,0.06777599950631459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,8,power_law_1.2,22.004275512695312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,16,1,balanced,1.5234346389770508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,16,1,power_law_1.01,0.040012800693511964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,16,1,power_law_1.01,0.04746879935264588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,8,1,power_law_1.2,9.709286499023438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,16,1,power_law_1.01,0.05255039930343628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,8,1,power_law_1.01,9.72943344116211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,16,1,power_law_1.01,0.05968639850616455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,16,1,power_law_1.01,0.03149439990520477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,16,1,power_law_1.01,0.08878080248832702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,16,1,power_law_1.01,0.04431999921798706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,16,1,power_law_1.01,0.09093760251998902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,16,1,power_law_1.01,0.04514560103416443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,16,1,power_law_1.01,0.09126399755477906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,16,1,power_law_1.01,0.0490880012512207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,16,1,power_law_1.01,0.09180160164833069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,8,power_law_1.2,19.66809539794922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,16,1,balanced,0.0693333347638448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,16,1,power_law_1.01,0.05102720260620117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,16,1,balanced,0.07161599894364674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,16,1,balanced,0.0703306645154953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,16,1,power_law_1.01,0.052211201190948485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,16,1,balanced,0.07261333366235097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,16,1,balanced,0.0737066666285197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,16,1,power_law_1.01,0.05312640070915222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,16,1,balanced,0.07261866827805837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,16,1,balanced,0.07875200112660725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,16,1,power_law_1.01,0.05192959904670715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,16,1,balanced,0.07753600180149078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,16,1,balanced,0.0846453309059143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,16,1,power_law_1.01,0.05243520140647888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,16,1,balanced,0.10170132915178935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,16,1,balanced,0.027215999861558277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,16,1,power_law_1.01,0.05575680136680603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,16,1,power_law_1.01,0.056831997632980344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,16,1,power_law_1.01,0.048895999789237976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,16,1,power_law_1.01,0.06058239936828613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,16,1,power_law_1.01,0.049235200881958006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,16,1,power_law_1.01,0.06993280053138733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,16,1,power_law_1.01,0.07385600209236146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,16,1,power_law_1.01,0.07656319737434387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,16,1,power_law_1.01,0.08428800106048584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,16,1,balanced,0.02757866680622101
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,16,1,power_law_1.01,0.02377600073814392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,16,1,balanced,0.030159999926884968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,16,1,power_law_1.01,0.08887680172920227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,16,1,balanced,0.02937600016593933
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,16,1,balanced,0.03153600047032038
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,16,1,balanced,0.03145066648721695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,16,1,power_law_1.01,0.09492480158805847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,16,1,balanced,0.03403733422358831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,16,1,balanced,0.0340639998515447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,8,power_law_1.2,15.122023010253907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,16,1,balanced,0.035749333600203194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,16,1,balanced,0.03562133262554804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,16,1,power_law_1.01,0.051801598072052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,16,1,balanced,0.03674133370320002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,16,1,power_law_1.2,0.04479359984397888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,16,1,power_law_1.01,0.06502400040626526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,16,1,power_law_1.01,0.0936959981918335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,16,1,power_law_1.2,0.04850560128688812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,16,1,power_law_1.01,0.06698880195617676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,16,1,power_law_1.01,0.09532160162925721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,16,1,power_law_1.2,0.05097600221633911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,16,1,power_law_1.01,0.06908800005912781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,16,1,power_law_1.2,0.06151679754257202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,16,1,power_law_1.01,0.06992639899253845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,16,1,power_law_1.2,0.09028480052947999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,16,1,power_law_1.01,0.07160320281982421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,16,1,power_law_1.2,0.09010559916496277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,16,1,balanced,0.10301333665847778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,16,1,balanced,0.10185600320498149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,16,1,power_law_1.2,0.0927232027053833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,16,1,balanced,0.12518399953842163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,16,1,balanced,0.16318933169047037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,16,1,power_law_1.01,0.024255999922752382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,16,1,balanced,0.21308799584706625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,16,1,power_law_1.01,0.025446400046348572
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,16,1,power_law_1.01,0.02743679881095886
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,16,1,power_law_1.01,0.027878400683403016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,16,1,balanced,0.03991466760635376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,8,1,power_law_1.2,9.502649688720703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,16,1,balanced,0.04162666698296865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,16,1,power_law_1.01,0.03017599880695343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,16,1,balanced,0.04148799926042557
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,16,1,balanced,0.05416533350944519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,16,1,power_law_1.2,0.035148799419403076
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,16,1,balanced,0.05208533505598704
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,16,1,balanced,0.04594666759173075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,16,1,power_law_1.2,0.04012799859046936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,16,1,balanced,0.08338133494059245
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,16,1,balanced,0.06619200110435486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,16,1,power_law_1.2,0.04778240025043488
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,16,1,balanced,0.08262399832407634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,16,1,power_law_1.2,0.04896639883518219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,16,1,power_law_1.01,0.07274240255355835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,16,1,power_law_1.01,0.12616959810256959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,16,1,power_law_1.2,0.0664192020893097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,16,1,power_law_1.2,0.0670527994632721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,16,1,power_law_1.2,0.0939520001411438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,16,1,power_law_1.2,0.06828160285949707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,16,1,balanced,0.26417599121729535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,16,1,power_law_1.2,0.07010560035705567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,16,1,balanced,0.37174399693806964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,16,1,power_law_1.2,0.0719488024711609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,16,1,balanced,0.4787946542104085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,16,1,power_law_1.2,0.07118080258369446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,16,1,balanced,0.6945439974466959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,16,1,power_law_1.01,0.031641599535942075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,16,1,power_law_1.2,0.07747840285301208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,16,1,power_law_1.01,0.09779199957847595
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,16,1,power_law_1.01,0.033369600772857666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,16,1,power_law_1.2,0.08083840012550354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,16,1,power_law_1.01,0.1442304015159607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,16,1,power_law_1.01,0.03336319923400879
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,16,1,balanced,0.09752000371615092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,16,1,balanced,0.12614400188128153
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,16,1,power_law_1.01,0.03367680013179779
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,16,1,balanced,0.15927466750144958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,16,1,power_law_1.01,0.07819520235061646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,16,1,power_law_1.01,0.08262400031089782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,16,1,power_law_1.01,0.08672000169754028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,16,1,power_law_1.01,0.09363840222358703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,16,1,balanced,0.9126986662546793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,16,1,power_law_1.01,0.10113919973373413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,16,1,power_law_1.01,0.19541759490966798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,16,1,power_law_1.01,0.10238080024719239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,16,1,power_law_1.01,0.10800000429153442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,16,1,power_law_1.01,0.11411839723587036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,16,1,power_law_1.01,0.036262398958206175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,16,1,power_law_1.01,0.14766720533370972
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,16,1,power_law_1.01,0.0383103996515274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,16,1,power_law_1.2,0.09442560076713562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,16,1,power_law_1.01,0.1748095989227295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,16,1,power_law_1.2,0.09631360173225403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,16,1,power_law_1.01,0.03906559944152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,16,1,power_law_1.01,0.22802560329437255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,16,1,power_law_1.01,0.24688639640808105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,16,1,power_law_1.2,0.09879040122032165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,16,1,power_law_1.2,0.10351999998092651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,16,1,power_law_1.2,0.10673279762268066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,16,1,power_law_1.2,0.08291839957237243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,16,1,power_law_1.2,0.11191040277481079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,16,1,power_law_1.2,0.09004160165786743
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,16,1,balanced,0.2242400050163269
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,16,1,balanced,0.2919466694196065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,16,1,balanced,0.42156267166137695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,16,1,power_law_1.2,0.11893119812011718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,16,1,balanced,0.5456159909566244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,16,1,power_law_1.01,0.043315199017524716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,16,1,power_law_1.2,0.0991424024105072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,14336,2,8,16,1,power_law_1.2,0.042105600237846375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,14336,2,8,16,1,power_law_1.2,0.047628799080848695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,16,1,power_law_1.01,0.04373759925365448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,16,1,power_law_1.2,0.10911359786987304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,16,1,power_law_1.2,0.12677119970321654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,16,1,balanced,1.3444639841715496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,16,1,power_law_1.2,0.11367039680480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,16,1,power_law_1.01,0.04550400078296661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,16,1,balanced,1.7830026944478352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,16,1,power_law_1.01,0.04870400130748749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,16,1,power_law_1.2,0.12755839824676513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,16,1,power_law_1.2,0.14710400104522706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,14336,2,8,16,1,power_law_1.2,0.04917759895324707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,14336,2,8,16,1,power_law_1.2,0.05146880149841308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,14336,2,8,16,1,power_law_1.2,0.05261440277099609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,16,1,power_law_1.01,0.10474879741668701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,16,1,power_law_1.2,0.150271999835968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,16,1,power_law_1.01,0.2803584098815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,16,1,power_law_1.01,0.11161600351333618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,16,1,power_law_1.01,0.11760640144348145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,16,1,power_law_1.01,0.05985919833183288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,16,1,power_law_1.01,0.12735359668731688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,16,1,power_law_1.01,0.3327552080154419
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,16,1,power_law_1.2,0.026073598861694337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,16,1,power_law_1.01,0.06588159799575806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,16,1,power_law_1.2,0.17468160390853882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,16,1,power_law_1.2,0.026080000400543212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,16,1,power_law_1.01,0.38655359745025636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,16,1,power_law_1.2,0.02698880136013031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,14336,2,8,16,1,power_law_1.2,0.05126399993896484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,14336,2,8,16,1,power_law_1.2,0.053260797262191774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,14336,2,8,16,1,power_law_1.2,0.05324800014495849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,14336,2,8,16,1,power_law_1.2,0.05607680082321167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,14336,2,8,16,1,power_law_1.2,0.0561024010181427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,14336,2,8,16,1,power_law_1.2,0.060198402404785155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,14336,2,8,16,1,power_law_1.2,0.06755200028419495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,16,1,power_law_1.01,0.12732800245285034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,14336,2,8,16,1,power_law_1.2,0.06925439834594727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,16,1,power_law_1.2,0.18026880025863648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,16,1,power_law_1.01,0.08436480164527893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,14336,2,8,16,1,power_law_1.2,0.0732479989528656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,14336,2,8,16,1,power_law_1.2,0.08353279829025269
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,16,1,power_law_1.2,0.028825598955154418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,16,1,power_law_1.2,0.23237121105194092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,16,1,power_law_1.2,0.029120001196861266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,16,1,power_law_1.2,0.2262336015701294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,16,1,power_law_1.01,0.1491711974143982
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,16,1,power_law_1.01,0.09975039958953857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,16,1,power_law_1.01,0.18474240303039552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,14336,2,8,16,1,power_law_1.2,0.08550400137901307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,16,1,power_law_1.2,0.030995199084281923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,14336,2,8,16,1,power_law_1.2,0.09566079974174499
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,16,1,power_law_1.2,0.032716798782348636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,16,1,power_law_1.2,0.03377279937267304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,14336,2,8,16,1,power_law_1.2,0.12714240550994874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,16,1,power_law_1.2,0.03505280017852783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,16,1,power_law_1.2,0.28193280696868894
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,16,1,power_law_1.2,0.03511039912700653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,16,1,power_law_1.2,0.035519999265670774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,16,1,power_law_1.2,0.03768959939479828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,16,1,power_law_1.2,0.038899201154708865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,16,1,power_law_1.2,0.04206080138683319
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,16,1,power_law_1.2,0.04281600117683411
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,16,1,power_law_1.01,0.12778879404067994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,16,1,power_law_1.01,0.4966464042663574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,16,1,power_law_1.01,0.23695359230041504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,16,1,power_law_1.2,0.04552960097789764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,16,1,power_law_1.01,0.42709760665893554
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,16,1,power_law_1.2,0.0480320006608963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,14336,2,8,16,1,power_law_1.2,0.14414080381393432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,16,1,power_law_1.2,0.060761600732803345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,16,1,power_law_1.01,0.2887295961380005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,16,1,power_law_1.2,0.2907263994216919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,16,1,power_law_1.2,0.07029759883880615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,14336,2,8,16,1,power_law_1.2,0.195961594581604
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,16,1,power_law_1.2,0.08430079817771911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,16,1,power_law_1.01,0.39228799343109133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,16,1,power_law_1.2,0.3875711917877197
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,16,1,power_law_1.01,0.16229759454727172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,14336,2,8,16,1,power_law_1.2,0.24702079296112062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,16,1,power_law_1.2,0.09927679896354676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,16,1,power_law_1.2,0.12792320251464845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,16,1,power_law_1.2,0.3939199924468994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,14336,2,8,16,1,power_law_1.2,0.33004159927368165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,16,1,power_law_1.01,0.506060791015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,16,1,power_law_1.01,0.70796799659729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,16,1,power_law_1.01,0.6146240234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,16,1,power_law_1.2,0.1604864001274109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,16,1,power_law_1.2,0.49748477935791013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,16,1,power_law_1.2,0.49816322326660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,14336,2,8,16,1,power_law_1.2,0.4257472038269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,16,1,power_law_1.01,0.22739200592041015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,16,1,power_law_1.2,0.7096831798553467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,16,1,power_law_1.01,0.7140223979949951
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,16,1,power_law_1.2,0.22764160633087158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,16,1,power_law_1.01,0.7987775802612305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,16,1,power_law_1.2,0.7133312225341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,14336,2,8,16,1,power_law_1.2,0.6120960235595703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,16,1,power_law_1.01,0.9239359855651855
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,16,1,power_law_1.01,0.2953216075897217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,16,1,power_law_1.01,1.1753472328186034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,16,1,power_law_1.2,0.9213760375976563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,16,1,power_law_1.01,0.9318464279174805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,16,1,power_law_1.2,0.29682559967041017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,16,1,power_law_1.01,0.4226240158081055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,14336,2,8,16,1,power_law_1.2,0.7962240219116211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,16,1,power_law_1.01,1.3696640014648438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,16,1,power_law_1.2,0.4209343910217285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,16,1,power_law_1.01,1.3566720008850097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,16,1,power_law_1.2,0.9372480392456055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,16,1,power_law_1.01,1.5503616333007812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,16,1,balanced,0.03549866626660029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,16,1,balanced,0.07189333438873291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,16,1,balanced,0.08709333340326945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,16,1,balanced,0.08752533793449402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,16,1,balanced,0.1258026659488678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,16,1,balanced,0.1276746690273285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,16,1,balanced,0.12949867049853006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,16,1,balanced,0.12984533111254373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,16,1,balanced,0.130730668703715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,16,1,balanced,0.13051733374595642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,16,1,balanced,0.13289067149162292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,16,1,balanced,0.13596266508102417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,16,1,balanced,0.1409280002117157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,16,1,balanced,0.14441066980361938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,16,1,balanced,0.16545066237449646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,16,1,balanced,0.17006399234135947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,16,1,balanced,0.16317333777745566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,16,1,balanced,0.19237866004308066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,16,1,balanced,0.24951465924580893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,16,1,balanced,0.3399893442789714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,16,1,balanced,0.41342933972676593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,16,1,balanced,0.5847306648890177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,16,1,balanced,0.7594026724497477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,16,1,balanced,1.116223971048991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,16,1,balanced,1.4722827275594075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,16,1,balanced,2.1812000274658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,16,1,balanced,2.920703887939453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,16,1,balanced,0.03748800108830134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,16,1,balanced,0.06403733293215434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,16,1,balanced,0.07974933087825775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,16,1,balanced,0.08092799782752991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,16,1,balanced,0.0988159974416097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,16,1,balanced,0.09910933176676433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,16,1,balanced,0.10053867101669312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,16,1,balanced,0.103301336367925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,16,1,balanced,0.10347732901573181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,16,1,balanced,0.10393599669138591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,16,1,balanced,0.10771200060844421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,16,1,balanced,0.10971200466156006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,16,1,balanced,0.11743467052777608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,16,1,balanced,0.12164800365765889
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,16,1,power_law_1.01,0.5499263763427734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,16,1,balanced,0.04782933493455251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,16,1,balanced,0.06021333237489065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,16,1,balanced,0.07374933362007141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,16,1,balanced,0.07454399764537811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,16,1,balanced,0.07457600037256877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,16,1,balanced,0.07669866581757863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,16,1,balanced,0.08082666496435802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,16,1,balanced,0.0802400012811025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,16,1,balanced,0.08094933132330577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,16,1,balanced,0.08266133566697438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,16,1,balanced,0.08566400408744812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,16,1,balanced,0.09166933099428813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,16,1,balanced,0.09565333525339763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,16,1,balanced,0.10900266965230306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,16,1,balanced,0.11353600025177002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,16,1,balanced,0.11756267150243123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,16,1,balanced,0.12123200297355652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,16,1,balanced,0.17187732458114624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,16,1,balanced,0.19688532749811807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,16,1,balanced,0.1388746698697408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,16,1,balanced,0.2759840091069539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,16,1,balanced,0.35225598017374676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,16,1,balanced,0.4944053490956624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,16,1,balanced,0.6311999956766764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,16,1,balanced,0.9254026412963867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,16,1,balanced,1.206976016362508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,16,1,balanced,1.7798506418863933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,16,1,balanced,0.14457066853841147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,16,1,balanced,0.13939733306566873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,16,1,balanced,0.19022933642069498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,16,1,balanced,0.23574932416280112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,16,1,balanced,2.348576068878174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,16,1,balanced,0.3251413305600484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,16,1,balanced,0.40223999818166095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,16,1,balanced,0.573194662729899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,16,1,balanced,0.7490293184916178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,16,1,balanced,1.1030133565266926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,16,1,balanced,1.4591093063354492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,16,1,balanced,2.166447957356771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,16,1,balanced,2.871232032775879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,16,1,power_law_1.01,0.059680002927780154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,16,1,power_law_1.01,0.06855679750442505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,16,1,power_law_1.01,0.07563520073890687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,16,1,power_law_1.01,0.08598399758338929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,16,1,power_law_1.01,0.12297600507736206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,16,1,power_law_1.01,0.12487679719924927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,16,1,balanced,0.026933332284291584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,16,1,balanced,0.027482666075229645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,16,1,power_law_1.01,0.12556159496307373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,16,1,balanced,0.03229333211978277
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,16,1,balanced,0.03268266717592875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,16,1,power_law_1.01,0.127455997467041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,16,1,balanced,0.03356799980004629
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,16,1,balanced,0.034714666505654655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,16,1,balanced,0.04214933514595032
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,16,1,balanced,0.04391466577847799
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,16,1,balanced,0.043840001026789345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,16,1,balanced,0.04433066646258036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,16,1,balanced,0.04599999884764353
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,16,1,balanced,0.05199466645717621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,16,1,balanced,0.05309866865475973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,16,1,balanced,0.05600533386071523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,16,1,balanced,0.06983466446399689
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,16,1,balanced,0.06409066418806712
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,16,1,balanced,0.0641599992911021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,16,1,balanced,0.09877333045005798
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,16,1,balanced,0.08460799853006999
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,16,1,balanced,0.10527466734250386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,16,1,balanced,0.12891200184822083
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,16,1,balanced,0.17894399166107178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,16,1,balanced,0.22383999824523926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,16,1,balanced,0.32413333654403687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,16,1,power_law_1.01,0.1299008011817932
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,16,1,balanced,0.4166933298110962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,16,1,power_law_1.01,0.1313088059425354
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,16,1,balanced,0.6022080183029175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,16,1,power_law_1.01,0.1349120020866394
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,16,1,balanced,0.7867253621419271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,16,1,power_law_1.01,0.14781440496444703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,16,1,power_law_1.01,0.05343359708786011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,16,1,power_law_1.01,0.06228479743003845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,16,1,power_law_1.01,0.15074559450149536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,16,1,power_law_1.01,0.06717439889907836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,16,1,power_law_1.01,0.15555200576782227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,16,1,power_law_1.01,0.07541120052337646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,16,1,power_law_1.01,0.09468799829483032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,16,1,power_law_1.01,0.09654399752616882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,16,1,power_law_1.01,0.0989247977733612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,16,1,power_law_1.01,0.10016640424728393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,16,1,power_law_1.01,0.1045375943183899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,16,1,power_law_1.01,0.10686719417572021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,16,1,power_law_1.01,0.11100800037384033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,16,1,power_law_1.01,0.11666560173034668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,16,1,power_law_1.01,0.12100479602813721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,16,1,power_law_1.01,0.12952959537506104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,16,1,power_law_1.01,1.7986047744750977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,16,1,power_law_1.01,0.1361791968345642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,16,1,power_law_1.01,0.049932798743247984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,16,1,power_law_1.01,0.059628802537918094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,16,1,power_law_1.01,0.0633408010005951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,16,1,power_law_1.01,0.15675519704818724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,16,1,power_law_1.2,1.3612799644470215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,16,1,power_law_1.01,0.07368959784507752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,16,1,power_law_1.01,0.07286400198936463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,16,1,power_law_1.01,0.17102080583572388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,16,1,power_law_1.01,0.07396479845046997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,16,1,power_law_1.01,0.07835519909858704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,16,1,power_law_1.01,0.07902079820632935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,16,1,power_law_1.01,0.08094080090522766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,16,1,power_law_1.01,0.16335999965667725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,16,1,power_law_1.01,0.21534080505371095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,16,1,power_law_1.01,0.08095999956130981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,16,1,power_law_1.01,0.2607680082321167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,16,1,power_law_1.01,0.17115520238876342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,16,1,power_law_1.2,1.3621503829956054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,16,1,power_law_1.01,0.3473151922225952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,16,1,power_law_1.01,0.17821439504623413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,16,1,power_law_1.01,0.23790080547332765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,16,1,power_law_1.01,0.08384000062942505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,16,1,power_law_1.01,0.08934400081634522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,14336,2,8,16,1,power_law_1.2,1.1696895599365233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,16,1,power_law_1.01,0.09164159893989562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,16,1,power_law_1.01,0.28743040561676025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,16,1,power_law_1.01,0.10042879581451417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,16,1,power_law_1.01,0.11151360273361206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,16,1,power_law_1.01,0.42914562225341796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,16,1,power_law_1.01,0.1155392050743103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,16,1,power_law_1.01,0.1435968041419983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,16,1,power_law_1.01,0.3636352062225342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,16,1,power_law_1.01,1.7976255416870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,16,1,power_law_1.01,0.4450943946838379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,16,1,power_law_1.01,0.17111680507659913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,16,1,power_law_1.2,1.8004928588867188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,16,1,power_law_1.2,0.05902720093727112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,16,1,power_law_1.2,0.06782079935073852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,16,1,power_law_1.2,0.07485439777374267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,16,1,power_law_1.01,0.6310783863067627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,16,1,power_law_1.2,0.08696320056915283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,16,1,power_law_1.2,0.1238976001739502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,16,1,power_law_1.01,0.1972607970237732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,16,1,power_law_1.2,0.12496639490127563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,16,1,power_law_1.2,0.12670079469680787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,16,1,power_law_1.2,0.12835839986801148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,16,1,power_law_1.01,0.27336320877075193
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,16,1,power_law_1.01,0.027027198672294618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,16,1,power_law_1.2,0.12926080226898193
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,16,1,power_law_1.01,0.029177600145339967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,16,1,power_law_1.2,0.13156479597091675
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,16,1,power_law_1.01,0.02885119915008545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,16,1,power_law_1.01,0.031763198971748355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,16,1,power_law_1.2,0.13648639917373656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,16,1,power_law_1.01,0.03329919874668121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,16,1,power_law_1.01,0.03797760009765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,16,1,power_law_1.01,0.3469887971878052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,16,1,power_law_1.01,0.04069760143756866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,16,1,power_law_1.01,0.0449535995721817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,16,1,power_law_1.01,0.6117887973785401
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,16,1,power_law_1.01,0.044268798828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,16,1,power_law_1.01,0.044889599084854126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,16,1,power_law_1.01,0.04983679950237274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,16,1,power_law_1.01,0.049721598625183105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,16,1,power_law_1.01,0.049958398938179015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,16,1,power_law_1.01,0.05913599729537964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,16,1,power_law_1.2,0.14931199550628663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,16,1,power_law_1.2,0.15114879608154297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,16,1,power_law_1.01,0.05845119953155518
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,16,1,power_law_1.01,0.06098560094833374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,16,1,power_law_1.2,0.1551743984222412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,16,1,power_law_1.2,0.16494079828262329
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,16,1,power_law_1.01,0.06835839748382569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,16,1,power_law_1.01,0.8024959564208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,16,1,power_law_1.01,0.07605119943618774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,16,1,power_law_1.01,0.09221119880676269
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,16,1,power_law_1.01,0.11214079856872558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,16,1,power_law_1.2,0.1737280011177063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,16,1,power_law_1.01,0.7895423889160156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,16,1,power_law_1.01,0.49621758460998533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,16,1,power_law_1.2,0.17639039754867553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,16,1,power_law_1.2,0.23160960674285888
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,16,1,power_law_1.01,0.13546240329742432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,16,1,power_law_1.2,0.2940864086151123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,16,1,power_law_1.01,0.637004804611206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,16,1,power_law_1.01,1.1600959777832032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,16,1,power_law_1.2,0.37845759391784667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,16,1,power_law_1.01,1.141113567352295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,16,1,power_law_1.2,0.44801921844482423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,16,1,power_law_1.01,0.18348159790039062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,16,1,power_law_1.01,0.9199551582336426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,16,1,power_law_1.2,0.6243584156036377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,16,1,power_law_1.2,0.5478847980499267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,16,1,power_law_1.2,0.058233600854873654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,16,1,power_law_1.2,0.058719998598098753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,16,1,power_law_1.2,0.0657151997089386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,16,1,power_law_1.2,0.0761023998260498
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,16,1,power_law_1.01,0.23066880702972412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,16,1,power_law_1.2,0.09535359740257263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,16,1,power_law_1.01,1.490982437133789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,16,1,power_law_1.2,0.09641600251197815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,16,1,power_law_1.2,0.09846400022506714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,16,1,power_law_1.2,0.100108802318573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,16,1,power_law_1.2,0.10451840162277222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,16,1,power_law_1.2,0.10740480422973633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,16,1,power_law_1.2,0.11166720390319824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,16,1,power_law_1.2,0.11822079420089722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,16,1,power_law_1.2,0.12209279537200927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,16,1,power_law_1.2,0.12870399951934813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,16,1,power_law_1.2,0.1365504026412964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,16,1,power_law_1.2,0.14871679544448851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,16,1,power_law_1.2,0.1686527967453003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,16,1,power_law_1.2,0.21587839126586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,16,1,power_law_1.01,1.2035200119018554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,16,1,power_law_1.2,0.26474239826202395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,16,1,power_law_1.01,1.5249855995178223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,16,1,power_law_1.2,0.800716781616211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,16,1,power_law_1.2,0.3405951976776123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,14336,2,8,16,1,power_law_1.2,1.5459839820861816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,16,1,power_law_1.2,0.05431680083274841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,16,1,power_law_1.2,0.05766400098800659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,16,1,power_law_1.2,0.06179839968681335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,16,1,power_law_1.2,0.07328640222549439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,16,1,power_law_1.2,0.07366399765014649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,16,1,power_law_1.2,0.07359359860420227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,16,1,power_law_1.2,0.4357952117919922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,16,1,power_law_1.2,0.07848320007324219
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,16,1,power_law_1.01,0.32674560546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,16,1,power_law_1.2,0.07900800108909607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,16,1,power_law_1.2,0.08011519908905029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,16,1,power_law_1.2,0.080595201253891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,16,1,power_law_1.2,0.08288000226020813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,16,1,power_law_1.2,0.08826239705085755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,16,1,power_law_1.2,0.09074559807777405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,16,1,power_law_1.2,0.10032639503479004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,16,1,power_law_1.2,0.11000959873199463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,16,1,power_law_1.2,0.1161344051361084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,16,1,power_law_1.2,0.14254080057144164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,16,1,power_law_1.2,1.1607680320739746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,16,1,power_law_1.01,0.4228544235229492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,16,1,power_law_1.2,0.16998399496078492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,16,1,power_law_1.2,0.6091263771057129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,16,1,power_law_1.2,0.19715839624404907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,16,1,power_law_1.2,0.27199358940124513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,16,1,power_law_1.01,2.2259328842163084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,16,1,power_law_1.2,0.3455359935760498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,16,1,power_law_1.2,1.8033536911010741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,16,1,power_law_1.2,1.5073023796081544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,16,1,power_law_1.2,0.7874303817749023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,16,1,power_law_1.2,0.02622720003128052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,16,1,power_law_1.2,0.027820798754692077
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,16,1,power_law_1.2,0.028972798585891725
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,16,1,power_law_1.2,0.03187839984893799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,16,1,power_law_1.01,2.202239990234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,16,1,power_law_1.2,0.03407360017299652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,16,1,power_law_1.2,0.037452799081802365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,16,1,power_law_1.2,0.03966720104217529
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,16,1,power_law_1.2,0.04490880072116852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,16,1,power_law_1.2,0.04248960018157959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,16,1,power_law_1.2,0.044249600172042845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,16,1,power_law_1.2,0.048582398891448976
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,16,1,power_law_1.2,0.048307201266288756
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,16,1,power_law_1.2,0.05066879987716675
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,16,1,power_law_1.2,0.06074240207672119
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,16,1,power_law_1.2,0.05861120223999024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,16,1,power_law_1.2,0.05925760269165039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,16,1,power_law_1.2,0.0679423987865448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,16,1,power_law_1.01,1.7711488723754882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,16,1,power_law_1.2,0.4954239845275879
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,16,1,power_law_1.01,0.6081920146942139
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,16,1,power_law_1.2,0.0796288013458252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,16,1,power_law_1.2,0.09189119935035706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,16,1,power_law_1.2,0.11478400230407715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,16,1,power_law_1.2,0.6331711769104004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,16,1,power_law_1.2,1.1309568405151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,16,1,power_law_1.2,0.13029119968414307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,16,1,power_law_1.2,0.17960959672927856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,16,1,power_law_1.01,2.9380863189697264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,16,balanced,0.07698666552702586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,16,balanced,0.0780266672372818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,16,balanced,0.09002133210500081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,16,balanced,0.10756267110506694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,16,balanced,0.16736533244450888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,16,balanced,0.3945866823196411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,16,balanced,0.3972906668980916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,16,balanced,0.3996640046437581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,16,balanced,0.40485866864522296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,16,balanced,0.40955201784769696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,16,balanced,0.4174079895019531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,16,balanced,0.42397332191467285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,16,balanced,0.43189334869384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,16,balanced,0.44574399789174396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,16,balanced,0.46317867437998456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,16,balanced,0.4764266808827718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,16,balanced,0.5096799929936727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,16,balanced,0.5574026505152384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,16,balanced,0.6154026587804159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,16,balanced,0.7638346354166666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,16,balanced,0.9102346897125244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,16,balanced,1.2567253112792969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,16,balanced,1.5269920031229656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,16,balanced,2.1663947105407715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,16,balanced,2.827114741007487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,16,balanced,4.35368537902832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,16,1,power_law_1.2,0.229913592338562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,16,balanced,6.328992207845052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,16,balanced,0.04643199841181437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,16,balanced,0.05264533559481303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,16,1,power_law_1.2,2.2275583267211916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,16,1,power_law_1.2,0.9129664421081543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,16,balanced,0.07597866654396057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,16,balanced,0.09471999605496724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,16,balanced,0.13537599643071493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,16,balanced,0.3217173417409261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,16,balanced,0.3264426589012146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,16,balanced,0.33053332567214966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,16,balanced,0.33531200885772705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,16,balanced,0.34113065401713055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,16,balanced,0.3505706787109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,16,balanced,0.35979199409484863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,16,balanced,0.370037317276001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,16,balanced,0.3859413464864095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,16,balanced,0.4061973492304484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,16,balanced,0.42539199193318683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,16,balanced,0.4602400064468384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,16,balanced,0.533514658610026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,16,balanced,0.6044426759084066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,16,balanced,0.7527946631113688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,16,balanced,0.8949600060780843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,16,balanced,1.2581653594970703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,16,balanced,1.5370292663574219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,16,balanced,2.1872000694274902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,16,balanced,2.8320372899373374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,16,balanced,4.124896049499512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,16,1,power_law_1.2,0.32794239521026614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,16,balanced,5.492181142171224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,16,balanced,0.0706826647122701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,16,balanced,0.07464000085989635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,16,balanced,0.09062932928403218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,16,balanced,0.10254399975140889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,16,balanced,0.14244266351064047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,16,balanced,0.26311999559402466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,16,balanced,0.2679786682128906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,16,balanced,0.27316800753275555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,16,balanced,0.2771093249320984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,16,balanced,0.2824160059293111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,16,balanced,0.28998400767644245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,16,balanced,0.2977013389269511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,16,balanced,0.30777599414189655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,16,balanced,0.32204800844192505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,16,balanced,0.3418826659520467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,16,balanced,0.3559466600418091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,16,balanced,0.38929065068562824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,16,balanced,0.4512480099995931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,16,balanced,0.5096640189488729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,16,balanced,0.629477341969808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,16,balanced,0.7577866713205973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,16,balanced,1.0707626342773438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,16,1,power_law_1.2,1.493734359741211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,16,balanced,1.3190613587697346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,16,balanced,1.8785227139790852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,16,balanced,2.4462560017903647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,16,balanced,3.5753866831461587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,16,balanced,4.704373359680176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,16,1,power_law_1.01,2.9107391357421877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,16,power_law_1.01,0.15166079998016357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,16,power_law_1.01,0.2993472099304199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,16,power_law_1.01,0.3066368103027344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,16,1,power_law_1.2,1.1969599723815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,16,power_law_1.01,0.30088961124420166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,16,power_law_1.01,0.28589439392089844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,16,power_law_1.01,0.3324352025985718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,16,power_law_1.01,0.3703167915344238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,16,balanced,0.029994666576385498
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,16,balanced,0.03148266673088074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,16,balanced,0.03159466634194056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,16,balanced,0.03766400118668874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,16,balanced,0.06131199995676676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,16,balanced,0.091648002465566
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,16,balanced,0.09438400467236836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,16,balanced,0.09597866733868916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,16,balanced,0.09612266222635905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,16,balanced,0.09694400429725647
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,16,balanced,0.0981226662794749
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,16,balanced,0.1340000033378601
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,16,balanced,0.13515200217564902
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,16,balanced,0.12612266341845194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,16,balanced,0.2869973381360372
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,16,balanced,0.2446026603380839
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,16,balanced,0.17972266674041748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,16,power_law_1.01,0.3768640041351318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,16,power_law_1.01,0.3902719974517822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,16,power_law_1.01,0.3820159912109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,16,1,power_law_1.2,2.932262420654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,16,power_law_1.01,0.37889919281005857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,16,power_law_1.01,0.11759999990463257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,16,power_law_1.01,0.41361279487609864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,16,power_law_1.01,0.19848320484161378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,16,power_law_1.01,0.41699838638305664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,16,balanced,0.13793067137400308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,16,power_law_1.01,0.1997760057449341
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,16,balanced,0.1442453364531199
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,16,balanced,0.16302399833997092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,16,power_law_1.01,0.2013375997543335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,16,balanced,0.18292800585428873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,16,balanced,0.22393065690994263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,16,power_law_1.01,0.20008320808410646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,16,balanced,0.2675146659215291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,16,power_law_1.01,0.2891135931015015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,16,power_law_1.01,0.31271040439605713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,16,power_law_1.01,0.3234112024307251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,16,power_law_1.01,0.4307712078094482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,16,power_law_1.01,0.45688958168029786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,16,power_law_1.01,0.483948802947998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,16,balanced,0.3798826535542806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,16,power_law_1.01,0.5275519847869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,16,balanced,0.4638986587524414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,16,balanced,0.6645706494649252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,16,power_law_1.01,0.6524479866027832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,16,balanced,0.8675519625345866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,16,power_law_1.01,0.324288010597229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,16,power_law_1.01,0.3301440000534058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,16,power_law_1.01,0.10595200061798096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,16,power_law_1.01,0.34180479049682616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,16,power_law_1.01,0.18163199424743653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,16,power_law_1.01,0.3513407945632935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,16,power_law_1.01,0.18388479948043823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,16,power_law_1.01,0.36700799465179446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,16,power_law_1.01,0.18630399703979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,16,power_law_1.01,0.3778048038482666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,16,power_law_1.01,0.19481600522994996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,16,power_law_1.01,0.7213888168334961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,16,power_law_1.01,0.39876480102539064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,16,power_law_1.01,0.43283839225769044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,16,power_law_1.01,0.8820672035217285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,16,power_law_1.01,0.24744319915771484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,16,power_law_1.01,0.5040319919586181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,16,power_law_1.01,1.0861568450927734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,16,power_law_1.01,0.2560256004333496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,16,power_law_1.01,0.24546558856964112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,16,power_law_1.01,1.4692543983459472
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,16,1,power_law_1.01,0.7903615951538085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,16,power_law_1.01,1.8933759689331056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,16,power_law_1.01,0.5962175846099853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,16,power_law_1.01,0.6888000011444092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,16,power_law_1.01,2.7238399505615236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,16,power_law_1.01,0.2473599910736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,16,power_law_1.01,0.9058624267578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,16,1,power_law_1.01,2.338617515563965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,16,power_law_1.2,0.15196160078048707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,16,power_law_1.01,1.0873727798461914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,16,power_law_1.2,0.280947208404541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,16,power_law_1.01,3.5461761474609377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,16,power_law_1.2,0.2272320032119751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,16,power_law_1.01,1.4805248260498047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,16,power_law_1.2,0.27948799133300783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,16,power_law_1.2,0.2965375900268555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,16,power_law_1.01,1.871392059326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,16,power_law_1.01,0.2600255966186523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,16,power_law_1.2,0.3607104063034058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,16,power_law_1.01,0.26416640281677245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,16,power_law_1.2,0.3722879886627197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,16,power_law_1.01,5.011769485473633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,16,power_law_1.2,0.3623231887817383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,16,power_law_1.01,2.6775360107421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,16,power_law_1.2,0.382860803604126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,16,power_law_1.2,0.3785599946975708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,16,power_law_1.01,0.056415998935699464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,16,power_law_1.2,0.3895616054534912
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,16,power_law_1.01,0.08458880186080933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,16,power_law_1.2,0.4045119762420654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,16,power_law_1.01,3.4949569702148438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,16,power_law_1.01,0.08357759714126586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,16,power_law_1.01,0.28614399433135984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,16,power_law_1.2,0.41403517723083494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,16,power_law_1.01,0.08344320058822632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,16,power_law_1.01,0.2906559944152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,16,power_law_1.2,0.42984957695007325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,16,power_law_1.01,6.941049957275391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,16,power_law_1.01,0.08681600093841553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,16,power_law_1.01,0.3091455936431885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,16,power_law_1.2,0.4472640037536621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,16,power_law_1.2,0.11596159934997559
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,16,power_law_1.01,0.07275519967079162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,16,power_law_1.01,0.325164794921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,16,power_law_1.2,0.5144896030426025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,16,power_law_1.2,0.19856640100479125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,16,power_law_1.01,0.0843455970287323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,16,power_law_1.2,0.5825535774230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,16,power_law_1.01,4.9418689727783205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,16,power_law_1.01,0.09009919762611389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,16,power_law_1.2,0.6532288074493409
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,16,power_law_1.01,0.08821759819984436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,16,power_law_1.2,0.7436927795410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,16,1,power_law_1.2,2.1948543548583985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,16,power_law_1.01,0.09137920141220093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,16,power_law_1.01,0.08471680283546448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,16,power_law_1.2,0.9393600463867188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,16,power_law_1.01,0.10892159938812256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,16,power_law_1.01,0.11119999885559081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,16,power_law_1.01,0.35143680572509767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,16,power_law_1.2,1.182252788543701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,16,power_law_1.01,0.10858880281448365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,16,power_law_1.01,0.3973184108734131
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,16,power_law_1.01,0.23176319599151612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,16,power_law_1.01,6.8358909606933596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,16,power_law_1.01,0.4732800006866455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,16,power_law_1.2,1.5555264472961425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,16,power_law_1.01,0.17690880298614503
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,16,power_law_1.01,0.15011199712753295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,16,power_law_1.2,0.10799360275268555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,16,power_law_1.01,0.13370239734649658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,16,power_law_1.2,0.1808832049369812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,16,power_law_1.2,2.027155113220215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,16,power_law_1.01,0.14608639478683472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,16,power_law_1.2,0.17596800327301027
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,16,power_law_1.01,0.1699072003364563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,16,power_law_1.2,0.17672959566116334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,16,power_law_1.2,0.1865407943725586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,16,1,power_law_1.2,0.42327041625976564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,16,power_law_1.01,0.19733760356903077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,16,power_law_1.2,0.1977280020713806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,16,power_law_1.2,0.20578560829162598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,16,power_law_1.2,2.936787223815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,16,power_law_1.2,0.2035968065261841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,16,power_law_1.2,0.23413119316101075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,16,power_law_1.01,0.5512959957122803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,16,power_law_1.2,0.2893631935119629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,16,power_law_1.2,0.2380608081817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,16,power_law_1.01,0.7049536228179931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,16,power_law_1.2,0.25063679218292234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,16,power_law_1.2,0.25382399559020996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,16,power_law_1.01,0.8793215751647949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,16,power_law_1.2,0.25418241024017335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,16,power_law_1.2,0.2762048006057739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,16,power_law_1.2,0.27589759826660154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,16,power_law_1.2,0.3033663988113403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,16,power_law_1.2,0.30833280086517334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,16,power_law_1.2,0.30164480209350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,16,power_law_1.2,0.32977919578552245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,16,power_law_1.2,0.31905279159545896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,16,power_law_1.2,3.800851058959961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,16,power_law_1.2,0.3627392053604126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,16,power_law_1.2,0.32866559028625486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,16,power_law_1.2,0.40119037628173826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,16,power_law_1.01,0.25447680950164797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,16,power_law_1.2,0.48919677734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,16,power_law_1.01,0.3195456027984619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,16,power_law_1.2,0.5574207782745362
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,16,power_law_1.01,0.44644479751586913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,16,power_law_1.2,0.723686408996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,16,power_law_1.2,5.540921783447265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,16,power_law_1.2,0.912019157409668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,16,power_law_1.2,0.33213438987731936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,16,power_law_1.2,0.3391551971435547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,16,power_law_1.2,1.2646464347839355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,16,power_law_1.2,0.35596799850463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,16,power_law_1.2,0.3567424058914185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,16,power_law_1.2,1.5598527908325195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,16,power_law_1.01,0.567577600479126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,16,power_law_1.2,0.3781183958053589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,16,power_law_1.2,0.43045759201049805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,16,power_law_1.01,0.8416959762573242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,16,power_law_1.01,1.1835968017578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,16,power_law_1.2,8.035526275634766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,16,power_law_1.2,0.452953577041626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,16,power_law_1.2,2.3376384735107423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,16,power_law_1.2,0.5438144207000732
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,16,power_law_1.01,1.1599488258361816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,16,power_law_1.01,1.5676095962524415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,16,1,balanced,0.036303999523321785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,16,1,balanced,0.06645333270231883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,16,1,balanced,0.08051200211048126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,16,1,balanced,0.10530666510264079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,16,1,balanced,0.16582399606704712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,16,1,balanced,0.34280534585316974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,16,power_law_1.2,2.941094398498535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,16,1,balanced,0.3454773426055908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,16,1,balanced,0.3403466542561849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,16,1,balanced,0.34206398328145343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,16,1,balanced,0.3471519947052002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,16,power_law_1.01,2.188768005371094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,16,1,balanced,0.34676798184712726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,16,1,balanced,0.3527413209279378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,16,1,balanced,0.3482666810353597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,16,1,balanced,0.3614826599756877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,16,1,balanced,0.4109813372294108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,16,1,balanced,0.4201120138168335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,16,power_law_1.2,4.414310455322266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,16,1,balanced,0.4362879991531372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,16,power_law_1.01,2.936966323852539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,16,1,balanced,0.4665439923604329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,16,1,balanced,0.49375998973846436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,16,1,balanced,0.5682880083719889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,16,1,balanced,0.6300693353017172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,16,1,balanced,0.9071946938832601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,16,1,balanced,1.0047199726104736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,16,power_law_1.01,4.305702209472656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,16,1,balanced,1.4246026674906414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,16,power_law_1.2,6.272844696044922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,16,1,balanced,0.04171733558177948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,16,1,balanced,0.06429333488146464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,16,1,balanced,1.825658639272054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,16,1,balanced,0.07650133470694225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,16,power_law_1.2,0.056831997632980344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,16,1,balanced,0.09899200002352397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,16,1,balanced,0.14669332901636759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,16,1,balanced,0.2664693395296733
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,16,power_law_1.2,0.08402559757232667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,16,1,balanced,0.2666613260904948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,16,1,balanced,0.2693546613057454
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,16,power_law_1.2,0.07687039971351624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,16,1,balanced,2.6061174074808755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,16,power_law_1.2,0.08320639729499817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,16,power_law_1.01,5.545196914672852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,16,power_law_1.2,0.08677120208740234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,16,power_law_1.2,0.6165056228637695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,16,power_law_1.2,0.0753279983997345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,16,power_law_1.2,0.08176000118255615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,16,power_law_1.2,0.7130623817443847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,16,power_law_1.2,0.08599680066108703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,16,power_law_1.2,0.8934656143188476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,16,1,balanced,0.27134400606155396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,16,power_law_1.2,0.08865919709205627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,16,1,balanced,0.2730720043182373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,16,1,balanced,0.27755733331044513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,16,power_law_1.2,0.0884607970714569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,16,1,balanced,0.2805546720822652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,16,1,balanced,0.28545065720876056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,16,power_law_1.2,1.1388480186462402
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,16,power_law_1.2,0.09041919708251953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,16,1,balanced,0.293232003847758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,16,1,balanced,0.34357333183288574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,16,power_law_1.2,0.1128767967224121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,16,1,balanced,3.685450553894043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,16,power_law_1.2,0.10952960252761841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,16,power_law_1.2,1.560870361328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,16,power_law_1.2,0.10695680379867553
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,16,power_law_1.2,0.21704959869384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,16,power_law_1.2,0.1844928026199341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,16,power_law_1.2,2.1000896453857423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,16,power_law_1.2,0.17706880569458008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,16,power_law_1.2,0.13722879886627198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,16,1,balanced,0.36476266384124756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,16,1,balanced,0.3837013244628906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,16,power_law_1.2,0.15237120389938355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,16,1,balanced,0.4190239906311035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,16,power_law_1.2,2.89752311706543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,16,1,balanced,0.4460586706797282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,16,power_law_1.2,0.17297919988632202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,16,1,balanced,0.5446293354034424
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,16,power_law_1.2,0.21541121006011962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,16,1,balanced,0.6060853401819865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,16,1,balanced,0.8510346412658691
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,16,power_law_1.2,0.27781760692596436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,16,1,balanced,0.9964586893717448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,16,power_law_1.2,0.35394558906555174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,16,power_law_1.2,3.7389247894287108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,16,1,balanced,1.3857332865397136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,16,power_law_1.2,0.4831679821014404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,16,1,balanced,1.7845546404520671
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,16,power_law_1.2,0.6530752182006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,16,1,balanced,0.030202666918436687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,16,power_law_1.2,0.9668671607971191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,16,1,balanced,2.5633813540140786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,16,power_law_1.2,5.711289596557617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,16,power_law_1.2,1.4166208267211915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,16,1,balanced,3.3626187642415366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,16,1,power_law_1.01,0.03493120074272156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,16,1,power_law_1.01,0.03933440148830414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,16,1,power_law_1.01,0.06167680025100708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,16,1,power_law_1.01,0.04647679924964905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,16,1,power_law_1.01,0.07909119725227357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,16,1,power_law_1.2,1.7715263366699219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,16,1,power_law_1.01,0.10357760190963745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,16,1,power_law_1.01,0.14086400270462035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,16,1,power_law_1.01,0.1871359944343567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,16,1,power_law_1.01,0.23663361072540284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,16,power_law_1.2,7.6378173828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,16,1,power_law_1.01,0.24734721183776856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,16,1,power_law_1.01,0.2518784046173096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,16,1,power_law_1.01,0.2580287933349609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,16,1,power_law_1.01,0.2783679962158203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,16,1,power_law_1.01,0.3051136016845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,16,1,power_law_1.01,0.3170687913894653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,16,1,power_law_1.01,0.342905592918396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,16,1,power_law_1.01,0.3891455888748169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,16,1,power_law_1.01,0.4009280204772949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,16,1,power_law_1.01,0.4246016025543213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,16,1,power_law_1.01,0.48679041862487793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,16,1,power_law_1.01,0.5723199844360352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,16,1,power_law_1.01,0.6415999889373779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,16,1,power_law_1.01,0.07489280104637146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,16,1,power_law_1.01,0.7227327823638916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,16,1,power_law_1.01,0.9233023643493652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,16,1,power_law_1.01,1.0957695960998535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,16,1,balanced,0.03223466624816259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,16,1,power_law_1.01,1.4878016471862794
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,16,1,balanced,0.035375999907652535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,16,1,balanced,0.047237331668535866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,16,1,power_law_1.01,0.0976256012916565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,16,1,power_law_1.01,0.12722560167312622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,16,1,power_law_1.01,1.8726335525512696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,16,1,power_law_1.01,0.1692031979560852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,16,1,power_law_1.01,0.20919039249420165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,16,1,power_law_1.01,0.22358400821685792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,16,1,power_law_1.01,2.653011131286621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,16,1,power_law_1.01,0.23525760173797608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,16,1,power_law_1.01,0.24440319538116456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,16,1,balanced,0.07454399764537811
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,16,1,balanced,0.1155413289864858
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,16,1,balanced,0.12167466680208842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,16,1,power_law_1.01,0.25000319480895994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,16,1,balanced,0.12406399846076965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,16,1,balanced,0.12719466288884482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,16,1,power_law_1.01,0.2630079984664917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,16,1,balanced,0.12797866264979044
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,16,1,balanced,0.13319999972979227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,16,1,balanced,0.13769066333770752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,16,1,power_law_1.01,0.27740800380706787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,16,1,power_law_1.01,3.4401409149169924
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,16,1,balanced,0.14437333742777506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,16,1,balanced,0.15446933110555014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,16,1,power_law_1.01,0.29937920570373533
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,16,1,balanced,0.1799466609954834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,16,1,balanced,0.190175990263621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,16,1,balanced,0.20933334032694498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,16,1,power_law_1.01,0.34053759574890136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,16,1,balanced,0.21280533075332642
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,16,1,balanced,0.23849600553512573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,16,1,power_law_1.01,0.35151360034942625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,16,1,balanced,0.2637439966201782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,16,1,balanced,0.30163200696309406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,16,1,power_law_1.01,0.3942464113235474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,16,1,balanced,0.39613866806030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,16,1,power_law_1.01,0.44466562271118165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,16,1,power_law_1.01,0.5054143905639649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,16,1,power_law_1.01,0.5835968017578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,16,1,power_law_1.01,0.6881984233856201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,16,1,power_law_1.01,0.8828991889953614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,16,1,balanced,0.45559998353322345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,16,1,balanced,0.6346240043640137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,16,1,power_law_1.01,1.0656703948974608
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,16,1,balanced,0.6695840358734131
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,16,1,balanced,0.9558826287587484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,16,1,power_law_1.01,1.4496064186096191
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,16,1,balanced,1.2390027046203613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,16,1,power_law_1.01,0.029260799288749695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,16,1,power_law_1.2,0.03470079898834229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,16,1,power_law_1.01,0.030675199627876282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,16,1,power_law_1.01,1.841913604736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,16,1,power_law_1.2,0.6058112144470215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,16,1,power_law_1.2,0.06252800226211548
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,16,1,power_law_1.01,0.03480319976806641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,16,1,power_law_1.2,0.07994880080223084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,16,1,power_law_1.01,0.04793600142002106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,16,1,balanced,0.051925331354141235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,16,1,power_law_1.2,0.10268160104751586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,16,1,power_law_1.01,0.06506239771842956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,16,1,power_law_1.01,2.6185535430908202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,16,1,power_law_1.2,0.13173760175704957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,16,1,power_law_1.01,0.08636800050735474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,16,1,power_law_1.2,0.17182719707489014
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,16,1,power_law_1.01,0.10456960201263428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,16,1,power_law_1.2,0.2302720069885254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,16,1,power_law_1.01,0.11067520380020142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,16,1,power_law_1.2,0.24510719776153564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,16,1,power_law_1.01,0.11172480583190918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,16,1,power_law_1.01,3.3976959228515624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,16,1,power_law_1.2,0.24864640235900878
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,16,1,power_law_1.01,0.11913599967956542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,16,1,power_law_1.2,0.03912320137023926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,16,1,power_law_1.2,0.2558655977249146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,16,1,power_law_1.01,0.12497919797897339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,16,1,power_law_1.2,0.04663040041923523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,16,1,power_law_1.2,0.2923327922821045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,16,1,power_law_1.2,0.07676799893379212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,16,1,power_law_1.2,0.31303040981292723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,16,1,power_law_1.2,0.09641600251197815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,16,1,power_law_1.2,0.32561919689178465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,16,1,power_law_1.2,0.12066559791564942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,16,1,power_law_1.2,0.3459007978439331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,16,1,power_law_1.2,0.15771520137786865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,16,1,power_law_1.2,0.39384961128234863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,16,1,power_law_1.2,0.20790400505065917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,16,1,power_law_1.2,0.4064767837524414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,16,1,power_law_1.2,0.2188800096511841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,16,1,power_law_1.2,0.4580992221832275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,16,1,power_law_1.01,0.1309183955192566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,16,1,power_law_1.2,0.23802239894866944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,16,1,power_law_1.2,0.5210752010345459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,16,1,power_law_1.2,0.24247679710388184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,16,1,power_law_1.2,0.5747072219848632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,16,1,power_law_1.2,0.6468992233276367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,16,1,power_law_1.2,2.9271488189697266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,16,1,power_law_1.2,0.7295807838439942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,16,1,power_law_1.2,0.9453568458557129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,16,1,power_law_1.01,0.1425984025001526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,16,1,power_law_1.01,0.1572160005569458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,16,1,power_law_1.2,0.2546112060546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,16,1,power_law_1.2,1.1153727531433106
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,16,1,power_law_1.01,0.17182079553604127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,16,1,power_law_1.2,0.26732800006866453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,16,1,power_law_1.01,0.1843135952949524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,16,1,power_law_1.2,0.2820096015930176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,16,1,power_law_1.01,0.21464319229125978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,16,1,power_law_1.2,0.30490880012512206
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,16,1,power_law_1.01,0.2111743927001953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,16,1,balanced,0.06467199822266896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,16,1,balanced,0.08077333370844524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,16,1,power_law_1.2,0.34532480239868163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,16,1,balanced,0.11334400375684102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,16,1,power_law_1.01,0.2468735933303833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,16,1,power_law_1.2,0.36895999908447263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,16,1,power_law_1.01,0.2668544054031372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,16,1,power_law_1.2,0.4007872104644775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,16,1,power_law_1.01,0.3171328067779541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,16,1,power_law_1.2,0.47151360511779783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,16,1,power_law_1.01,0.3930880069732666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,16,1,power_law_1.2,0.5059584140777588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,16,1,power_law_1.2,1.5087231636047362
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,16,1,power_law_1.01,0.520147180557251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,16,1,power_law_1.2,0.5960256099700928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,16,1,power_law_1.01,0.567142391204834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,16,1,power_law_1.2,0.6916160106658935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,16,1,power_law_1.2,1.8858047485351563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,16,1,power_law_1.01,0.7151936054229736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,16,1,power_law_1.2,0.8850111961364746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,16,1,power_law_1.01,0.9939135551452637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,16,1,power_law_1.2,1.0797311782836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,16,1,power_law_1.2,2.668063926696777
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,16,1,power_law_1.01,1.3254207611083983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,16,1,power_law_1.2,1.4639616012573242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,16,1,power_law_1.2,3.464019012451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,16,1,power_law_1.2,1.8487167358398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,16,balanced,0.048895999789237976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,16,balanced,0.04914666712284088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,16,balanced,0.08288000027338664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,16,balanced,0.09673066933949788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,16,balanced,0.14035200079282126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,16,balanced,0.34512531757354736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,16,balanced,0.3498773177464803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,16,balanced,0.35150933265686035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,16,balanced,0.35633599758148193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,16,1,power_law_1.2,2.6295679092407225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,16,balanced,0.360645333925883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,16,balanced,0.3666773239771525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,16,balanced,0.374944011370341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,16,balanced,0.38060800234476727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,16,balanced,0.3940639893213908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,16,balanced,0.4098293383916219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,16,balanced,0.427509347597758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,16,balanced,0.4490933418273926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,16,1,power_law_1.2,3.412870407104492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,16,balanced,0.5122720003128052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,16,balanced,0.04814399778842926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,16,balanced,0.04814933240413666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,16,balanced,0.5702666838963827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,16,balanced,0.0581226646900177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,16,balanced,0.08733333150545756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,16,balanced,0.12110400199890137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,16,balanced,0.701141357421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,16,1,power_law_1.01,0.05047680139541626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,16,balanced,0.8292160034179688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,16,balanced,1.1307040055592854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,16,1,balanced,0.18445332845052084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,16,1,balanced,0.3345973491668701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,16,1,power_law_1.2,0.029548799991607665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,16,1,balanced,0.33694934844970703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,16,1,balanced,0.33900264898935956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,16,1,power_law_1.2,0.030822399258613586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,16,1,balanced,0.3397013346354167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,16,1,balanced,0.3448479970296224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,16,1,power_law_1.2,0.03547520041465759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,16,1,balanced,0.3470613161722819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,16,1,balanced,0.3516106605529785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,16,1,power_law_1.2,0.045049598813056944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,16,1,balanced,0.3577599922815959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,16,1,balanced,0.3637973467508952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,16,1,power_law_1.2,0.06263039708137512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,16,1,balanced,0.3733706474304199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,16,1,balanced,0.38151466846466064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,16,1,power_law_1.2,0.08280959725379944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,16,1,balanced,0.39352532227834064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,16,1,balanced,0.420576016108195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,16,1,power_law_1.2,0.1033087968826294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,16,balanced,1.37553071975708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,16,1,balanced,0.447215994199117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,16,1,power_law_1.2,0.1083840012550354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,16,1,balanced,0.5027626752853394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,16,1,balanced,0.5789653460184733
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,16,1,power_law_1.2,0.11185920238494873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,16,1,balanced,0.8662026723225912
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,16,1,power_law_1.2,0.11870720386505126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,16,balanced,0.2939466635386149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,16,balanced,0.29686933755874634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,16,1,balanced,0.9833760261535645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,16,1,power_law_1.2,0.12520960569381714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,16,balanced,0.3017279903093974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,16,balanced,0.3072426716486613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,16,1,power_law_1.2,0.13351680040359498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,16,balanced,0.30823999643325806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,16,1,balanced,1.3982133865356445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,16,balanced,0.31993599732716876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,16,1,power_law_1.2,0.14213759899139405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,16,balanced,0.3305013378461202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,16,1,power_law_1.2,0.15797120332717896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,16,1,balanced,1.7926559448242188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,16,balanced,0.33975998560587567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,16,1,power_law_1.2,0.17238399982452393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,16,balanced,0.35283199946085614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,16,balanced,1.952730655670166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,16,balanced,0.37134401003519696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,16,1,power_law_1.2,0.18819199800491332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,16,balanced,0.3879679838816325
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,16,1,power_law_1.2,0.21612799167633057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,16,balanced,2.545685291290283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,16,balanced,3.896554629007975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,16,1,balanced,2.624096075693766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,16,1,power_law_1.2,0.21416959762573243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,16,1,power_law_1.2,0.05198079943656921
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,16,1,power_law_1.2,0.251859188079834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,16,1,power_law_1.01,0.05836799740791321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,16,1,power_law_1.2,0.05920640230178833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,16,1,balanced,3.4482078552246094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,16,1,power_law_1.2,0.08209279775619507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,16,1,power_law_1.2,0.1094208002090454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,16,1,power_law_1.2,0.1524672031402588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,16,balanced,0.42187201976776123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,16,1,power_law_1.2,0.21025919914245605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,16,balanced,0.48609598477681476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,16,balanced,0.553274671236674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,16,1,power_law_1.2,0.28613760471343996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,16,balanced,0.6850612958272299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,16,balanced,5.288965225219727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,16,1,power_law_1.2,0.29713280200958253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,16,balanced,0.8128426869710287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,16,1,power_law_1.01,0.08186240196228027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,16,1,power_law_1.2,0.3023423910140991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,16,balanced,1.141599973042806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,16,1,power_law_1.01,0.11167360544204712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,16,1,power_law_1.2,0.3070784091949463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,16,balanced,1.3967893918355305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,16,1,power_law_1.2,0.31440000534057616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,16,1,power_law_1.2,0.3184000015258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,16,balanced,1.9800267219543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,16,1,power_law_1.2,0.32704639434814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,16,1,power_law_1.2,0.3397631883621216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,16,1,power_law_1.2,0.26825599670410155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,16,balanced,2.565674622853597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,16,1,power_law_1.2,0.35879039764404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,16,1,power_law_1.2,0.32125439643859866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,16,balanced,0.06344000001748402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,16,balanced,0.07049066821734111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,16,balanced,0.08242666721343994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,16,balanced,0.09438932935396831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,16,1,power_law_1.2,0.3966527938842773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,16,balanced,0.12987732887268066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,16,balanced,0.23634666204452515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,16,balanced,0.24140799045562744
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,16,1,power_law_1.2,0.48793601989746094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,16,balanced,3.729109446207682
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,16,1,power_law_1.2,0.5732992172241211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,16,1,power_law_1.2,0.7213247776031494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,16,balanced,4.993845303853353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,16,1,power_law_1.2,0.3654016017913818
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,16,1,power_law_1.2,1.0041919708251954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,16,power_law_1.01,0.12833280563354493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,16,balanced,0.029685333371162415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,16,balanced,0.03173333406448364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,16,balanced,0.2424586613972982
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,16,balanced,0.031541332602500916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,16,balanced,0.24836800495783487
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,16,balanced,0.035674666364987694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,16,balanced,0.2507839997609456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,16,1,power_law_1.2,1.3506624221801757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,16,balanced,0.2608533302942912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,16,balanced,0.26890132824579877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,16,balanced,0.2742453416188558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,16,power_law_1.01,0.10634880065917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,16,balanced,0.2905493378639221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,16,balanced,0.30773333708445233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,16,power_law_1.01,0.17655680179595948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,16,balanced,0.32197866837183636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,16,balanced,0.3529599905014038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,16,power_law_1.01,0.1735360026359558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,16,balanced,0.40883731842041016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,16,balanced,0.46625598271687824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,16,1,power_law_1.2,0.39434878826141356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,16,balanced,0.5788480043411255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,16,power_law_1.01,0.25502719879150393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,16,balanced,0.6936960220336914
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,16,balanced,0.05765333275000254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,16,balanced,0.08331733445326488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,16,power_law_1.01,0.25845119953155515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,16,balanced,0.9828426837921143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,16,power_law_1.01,0.2587519884109497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,16,balanced,1.2069653669993083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,16,power_law_1.01,0.25333759784698484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,16,balanced,1.7267626126607258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,16,power_law_1.01,0.31371519565582273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,16,power_law_1.01,0.17894400358200074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,16,power_law_1.01,0.3184448003768921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,16,power_law_1.01,0.17825280427932738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,16,1,power_law_1.2,0.43422718048095704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,16,balanced,2.2451252937316895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,16,power_law_1.01,0.33393919467926025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,16,power_law_1.01,0.2773439884185791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,16,balanced,0.0844053328037262
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,16,balanced,0.08678399523099263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,16,1,power_law_1.01,0.15699199438095093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,16,power_law_1.01,0.32707839012145995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,16,power_law_1.01,0.28330240249633787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,16,power_law_1.01,0.33934080600738525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,16,power_law_1.01,0.2974400043487549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,16,power_law_1.01,0.3512320041656494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,16,1,power_law_1.2,2.3314624786376954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,16,power_law_1.01,0.29317119121551516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,16,power_law_1.01,0.36053760051727296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,16,power_law_1.01,0.29541759490966796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,16,power_law_1.01,0.10061440467834473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,16,power_law_1.01,0.36823039054870604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,16,power_law_1.01,0.31024000644683836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,16,power_law_1.01,0.15727360248565675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,16,power_law_1.01,0.3166591882705688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,16,balanced,0.08769067128499348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,16,power_law_1.01,0.1564288020133972
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,16,balanced,0.0888320008913676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,16,balanced,0.08898666501045227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,16,balanced,3.281989415486654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,16,power_law_1.01,0.16124800443649293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,16,power_law_1.01,0.18182400465011597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,16,power_law_1.01,0.2212928056716919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,16,power_law_1.01,0.22424960136413574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,16,balanced,4.322890599568685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,16,power_law_1.01,0.37827839851379397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,16,power_law_1.01,0.32958719730377195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,16,1,power_law_1.01,0.21982719898223876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,16,balanced,0.12457066774368286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,16,balanced,0.12793599565823874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,16,balanced,0.11547733346621196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,16,1,power_law_1.01,0.2906303882598877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,16,balanced,0.2815093398094177
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,16,balanced,0.2155946691830953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,16,balanced,0.16714666287104288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,16,balanced,0.12587199608484903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,16,balanced,0.1295253336429596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,16,balanced,0.14682666460673013
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,16,balanced,0.1632426679134369
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,16,balanced,0.19748266537984213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,16,power_law_1.01,0.39100799560546873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,16,balanced,0.23643199602762857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,16,balanced,0.3357813358306885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,16,power_law_1.01,0.3497215986251831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,16,balanced,0.4091893434524536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,16,power_law_1.01,0.36900479793548585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,16,balanced,0.5823359886805216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,16,power_law_1.01,0.3882496118545532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,16,1,power_law_1.01,0.29774720668792726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,16,power_law_1.01,0.45530238151550295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,16,1,power_law_1.01,0.30100479125976565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,16,1,power_law_1.2,0.5029632091522217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,16,power_law_1.01,0.5471615791320801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,16,1,power_law_1.01,0.30714240074157717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,16,1,power_law_1.01,0.31178879737854004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,16,1,power_law_1.2,0.5933375835418702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,16,power_law_1.01,0.6188608169555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,16,power_law_1.01,0.4293824195861816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,16,1,power_law_1.01,0.3218816041946411
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,16,power_law_1.01,0.05130239725112915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,16,power_law_1.01,0.7912320137023926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,16,1,power_law_1.2,0.7051904201507568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,16,power_law_1.01,0.4797952175140381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,16,1,power_law_1.01,0.3279040098190308
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,16,power_law_1.01,0.07544959783554077
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,16,balanced,0.7595787048339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,16,1,power_law_1.01,0.33556480407714845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,16,power_law_1.01,0.9688511848449707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,16,1,power_law_1.2,0.8851200103759765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,16,power_law_1.01,0.07455360293388366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,16,power_law_1.2,0.12721279859542847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,16,1,power_law_1.01,0.35088000297546384
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,16,power_law_1.01,0.075135999917984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,16,power_law_1.2,0.24805760383605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,16,1,power_law_1.2,1.0862336158752441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,16,power_law_1.01,1.2754048347473144
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,16,power_law_1.01,0.07807999849319458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,16,power_law_1.2,0.233024001121521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,16,power_law_1.01,0.07128959894180298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,16,power_law_1.2,0.2603967905044556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,16,1,power_law_1.2,1.4974080085754395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,16,power_law_1.01,1.6910655975341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,16,power_law_1.2,0.23247361183166504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,16,power_law_1.01,0.5794112205505371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,16,power_law_1.2,0.2947776079177856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,16,1,power_law_1.2,1.9047744750976563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,16,power_law_1.01,0.641593599319458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,16,power_law_1.01,0.22929279804229735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,16,power_law_1.2,0.3259968042373657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,16,1,power_law_1.01,0.3601408004760742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,16,power_law_1.01,0.8004608154296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,16,power_law_1.01,0.23557760715484619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,16,power_law_1.2,0.33162240982055663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,16,1,power_law_1.01,0.38147199153900146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,16,power_law_1.01,0.23753600120544432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,16,power_law_1.2,0.33660159111022947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,16,1,power_law_1.01,0.4277503967285156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,16,power_law_1.01,0.07088000178337098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,16,power_law_1.01,0.24145278930664063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,16,power_law_1.2,0.33904640674591063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,16,1,power_law_1.01,0.4949120044708252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,16,power_law_1.2,0.33315839767456057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,16,1,power_law_1.01,0.5827775955200195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,16,power_law_1.2,0.36221439838409425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,16,power_law_1.01,2.3535680770874023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,16,1,power_law_1.01,0.6715263843536377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,16,power_law_1.2,0.3638463973999023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,16,power_law_1.2,0.3736191987991333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,16,1,power_law_1.01,0.8785216331481933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,16,power_law_1.01,1.0143424034118653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,16,power_law_1.2,0.439961576461792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,16,power_law_1.01,3.1208255767822264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,16,1,power_law_1.01,1.0833855628967286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,16,power_law_1.01,0.07627519965171814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,16,power_law_1.01,1.3885248184204102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,16,power_law_1.01,0.08087040185928344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,16,1,power_law_1.01,1.4932031631469727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,16,power_law_1.01,1.7273599624633789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,16,1,power_law_1.01,1.8882816314697266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,16,1,power_law_1.2,2.6989120483398437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,16,power_law_1.01,4.637055969238281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,16,power_law_1.2,0.45854721069335935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,16,power_law_1.01,0.25377280712127687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,16,power_law_1.01,0.26334080696105955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,16,power_law_1.2,0.5134848117828369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,16,1,power_law_1.01,2.688620758056641
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,16,power_law_1.01,0.07890560030937195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,16,power_law_1.2,0.5914624214172364
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,16,power_law_1.01,0.08620799779891967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,16,power_law_1.2,0.6841472148895263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,16,power_law_1.01,0.10419199466705323
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,16,power_law_1.01,0.10104960203170776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,16,power_law_1.01,5.835935974121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,16,power_law_1.2,0.9011903762817383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,16,power_law_1.01,0.0942143976688385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,16,power_law_1.01,2.414259147644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,16,power_law_1.2,0.10631040334701539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,16,power_law_1.01,0.1840127944946289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,16,power_law_1.2,1.0293375968933105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,16,power_law_1.2,0.16509439945220947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,16,power_law_1.01,0.2785536050796509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,16,power_law_1.2,0.16676479578018188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,16,power_law_1.01,0.28923521041870115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,16,power_law_1.2,1.4357695579528809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,16,power_law_1.2,0.17703039646148683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,16,power_law_1.01,3.1498687744140623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,16,power_law_1.01,0.31379199028015137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,16,power_law_1.2,0.17912960052490234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,16,power_law_1.01,0.36272640228271485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,16,power_law_1.2,1.8235647201538085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,16,power_law_1.2,0.2716991901397705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,16,power_law_1.01,0.4300352096557617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,16,1,power_law_1.01,3.4831169128417967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,16,power_law_1.01,0.5003520011901855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,16,power_law_1.2,0.10008959770202637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,16,power_law_1.01,0.6475135803222656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,16,power_law_1.01,4.385939025878907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,16,power_law_1.2,2.5082815170288084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,16,1,power_law_1.2,3.509747314453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,16,power_law_1.2,0.15727360248565675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,16,power_law_1.01,0.8277567863464356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,16,power_law_1.2,0.1597759962081909
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,16,power_law_1.2,0.1638975977897644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,16,power_law_1.01,1.1034624099731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,16,power_law_1.2,0.1825152039527893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,16,power_law_1.2,3.2884544372558593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,16,power_law_1.2,0.2891648054122925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,16,power_law_1.01,1.3824383735656738
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,16,power_law_1.01,0.1800063967704773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,16,power_law_1.2,0.28463358879089357
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,16,1,power_law_1.2,0.7948480129241944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,16,1,balanced,0.03349866718053818
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,16,power_law_1.01,0.11746560335159302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,16,power_law_1.2,0.2932735919952393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,16,power_law_1.01,1.9899583816528321
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,16,power_law_1.01,0.11601920127868652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,16,power_law_1.2,0.2981055974960327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,16,power_law_1.01,0.1328703999519348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,16,power_law_1.2,0.31169281005859373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,16,power_law_1.2,4.969369506835937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,16,power_law_1.01,0.15427839756011963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,16,power_law_1.2,0.31905279159545896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,16,power_law_1.01,2.5633472442626952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,16,power_law_1.2,0.3272511959075928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,16,power_law_1.2,0.18944640159606935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,16,power_law_1.2,0.3497663974761963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,16,power_law_1.2,0.21710081100463868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,16,power_law_1.01,6.173535919189453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,16,power_law_1.2,0.368012809753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,16,power_law_1.2,0.217574405670166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,16,1,balanced,0.0378506655494372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,16,1,balanced,0.041802664597829185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,16,1,balanced,0.06758933266003926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,16,1,balanced,0.08874133229255676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,16,power_law_1.2,0.2200767993927002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,16,1,balanced,0.12823466459910074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,16,1,balanced,0.2304640014966329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,16,power_law_1.01,3.784627151489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,16,1,balanced,0.2342080076535543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,16,1,balanced,0.23360000054041544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,16,1,balanced,0.23496532440185547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,16,1,balanced,0.2370026707649231
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,16,power_law_1.2,0.053497600555419925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,16,power_law_1.2,6.707571411132813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,16,1,balanced,0.23883734146753946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,16,1,balanced,0.24540799856185913
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,16,power_law_1.2,0.07489920258522034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,16,power_law_1.2,0.06809599995613098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,16,1,balanced,0.06019733349482218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,16,1,balanced,0.07216000060240428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,16,1,balanced,0.09285866220792134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,16,1,balanced,0.14819733301798502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,16,power_law_1.2,0.4120063781738281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,16,1,balanced,0.28566400210062665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,16,power_law_1.01,5.021638488769531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,16,1,balanced,0.28806400299072266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,16,1,balanced,0.2887253363927205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,16,power_law_1.2,0.45122561454772947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,16,1,balanced,0.29595200220743817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,16,1,balanced,0.292197326819102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,16,1,balanced,0.2961866656939189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,16,power_law_1.2,0.5608960151672363
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,16,power_law_1.01,0.17681920528411865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,16,1,balanced,0.24794133504231772
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,16,power_law_1.01,0.2273344039916992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,16,1,balanced,0.25325334072113037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,16,1,balanced,0.3009173274040222
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,16,power_law_1.01,0.281113600730896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,16,1,balanced,0.32014934221903485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,16,1,balanced,0.3349493344624837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,16,1,balanced,0.36578134695688885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,16,power_law_1.01,0.39223039150238037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,16,1,balanced,0.3943626483281453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,16,1,balanced,0.4747946659723918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,16,power_law_1.01,0.5109951972961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,16,1,balanced,0.5306880076726278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,16,1,balanced,0.7372159957885742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,16,power_law_1.01,0.7223040103912354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,16,1,balanced,0.8650293350219727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,16,1,balanced,1.2038880189259846
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,16,power_law_1.01,1.0378303527832031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,16,power_law_1.2,0.07599999904632568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,16,1,power_law_1.01,0.031251201033592226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,16,power_law_1.2,0.07557759881019592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,16,power_law_1.2,0.22954881191253662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,16,1,balanced,0.02924799919128418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,16,1,balanced,0.03126933425664902
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,16,1,balanced,0.033557333052158356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,16,1,balanced,0.04398933549722036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,16,power_law_1.2,0.2464384078979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,16,1,balanced,0.06845333178838094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,16,1,balanced,0.3004853328069051
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,16,1,balanced,0.10501866539319356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,16,1,balanced,0.10533333818117778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,16,power_law_1.2,0.2552128076553345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,16,1,balanced,0.11111467083295186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,16,power_law_1.2,0.6478975772857666
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,16,1,balanced,0.1128000020980835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,16,1,balanced,0.1146453320980072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,16,power_law_1.2,0.2604543924331665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,16,power_law_1.2,0.27742719650268555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,16,power_law_1.2,0.30560638904571535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,16,1,balanced,1.549455960591634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,16,power_law_1.2,0.3300544023513794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,16,1,power_law_1.01,0.03697279989719391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,16,power_law_1.2,0.07919359803199769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,16,1,power_law_1.01,0.07041919827461243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,16,power_law_1.2,0.3664191961288452
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,16,power_law_1.2,0.07553920149803162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,16,1,balanced,0.3015999992688497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,16,1,power_law_1.01,0.09096320271492005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,16,1,balanced,0.3061866760253906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,16,power_law_1.2,0.43258237838745117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,16,1,balanced,0.3573013146718343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,16,1,power_law_1.01,0.12373759746551513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,16,1,balanced,0.3635093371073405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,16,1,balanced,0.37566399574279785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,16,power_law_1.2,0.5247488021850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,16,power_law_1.2,0.8472640037536621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,16,1,power_law_1.01,0.15926400423049927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,16,1,balanced,0.4030826489130656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,16,1,balanced,0.43380268414815265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,16,1,power_law_1.01,0.20195839405059815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,16,power_law_1.2,0.6771008014678955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,16,1,balanced,0.48664534091949463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,16,power_law_1.2,0.986956787109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,16,1,balanced,0.5784800052642822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,16,1,power_law_1.01,0.20903680324554444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,16,power_law_1.2,0.8524607658386231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,16,1,power_law_1.01,0.2175679922103882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,16,1,balanced,2.225930690765381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,16,1,power_law_1.01,0.22036480903625488
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,16,power_law_1.2,0.07765120267868042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,16,power_law_1.2,1.1269439697265624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,16,1,power_law_1.01,0.24832000732421874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,16,1,balanced,0.11825066804885864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,16,1,balanced,0.12191999951998393
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,16,power_law_1.2,0.08598399758338929
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,16,1,balanced,0.1265120009581248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,16,1,power_law_1.01,0.26869759559631345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,16,1,balanced,0.13608533143997192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,16,1,balanced,2.920405387878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,16,power_law_1.2,0.08335360288619995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,16,power_law_1.2,1.4813311576843262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,16,1,power_law_1.01,0.03583999872207642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,16,1,power_law_1.01,0.0409855991601944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,16,1,balanced,0.770693302154541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,16,1,power_law_1.01,0.06656640172004699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,16,power_law_1.2,2.1757247924804686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,16,1,balanced,0.8726186752319336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,16,power_law_1.2,1.4177472114562988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,16,1,power_law_1.01,0.08614400029182434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,16,1,balanced,1.2287733554840088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,16,1,power_law_1.01,0.1133504033088684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,16,1,power_law_1.01,0.1431872010231018
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,16,1,balanced,0.1607039968172709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,16,1,balanced,1.5727465947469075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,16,power_law_1.2,1.829555130004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,16,1,balanced,0.167087992032369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,16,1,power_law_1.01,0.27779839038848875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,16,1,balanced,0.18481600284576416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,16,power_law_1.2,2.999123191833496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,16,1,balanced,0.18710400660832724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,16,1,power_law_1.01,0.29536640644073486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,16,1,balanced,2.2546614011128745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,16,1,power_law_1.01,0.33623039722442627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,16,power_law_1.2,2.601875114440918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,16,1,balanced,2.931935946146647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,16,power_law_1.2,4.221996688842774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,16,1,power_law_1.01,0.17250560522079467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,16,power_law_1.2,0.08250880241394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,16,power_law_1.2,3.348160171508789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,16,1,balanced,0.20971733331680298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,16,1,power_law_1.01,0.1962048053741455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,16,1,balanced,0.2302293380101522
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,16,power_law_1.2,0.09359359741210938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,16,1,balanced,0.2639840046564738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,16,1,power_law_1.01,0.20421760082244872
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,16,power_law_1.2,0.10279680490493774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,16,1,power_law_1.01,0.34547839164733884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,16,power_law_1.2,0.09510400295257568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,16,1,power_law_1.01,0.36690559387207033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,16,power_law_1.2,0.1919487953186035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,16,1,power_law_1.01,0.42247681617736815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,16,power_law_1.2,0.14935040473937988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,16,power_law_1.2,0.13338880538940429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,16,power_law_1.2,4.842387390136719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,16,power_law_1.2,0.1221951961517334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,16,power_law_1.2,0.13734400272369385
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,16,1,balanced,0.3456960121790568
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,16,power_law_1.2,0.15917439460754396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,16,1,balanced,0.4007840156555176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,16,1,balanced,0.5650666554768881
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,16,power_law_1.2,0.18574080467224122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,16,power_law_1.2,0.24717440605163574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,16,1,balanced,0.5916906595230103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,16,1,power_law_1.01,0.4955264091491699
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,16,1,balanced,0.8265546957651774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,16,power_law_1.2,6.592396545410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,16,1,power_law_1.01,0.552351999282837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,16,1,balanced,1.070757309595744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,16,1,power_law_1.01,0.6243391990661621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,16,1,power_law_1.01,0.21063680648803712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,16,1,power_law_1.01,0.21727359294891357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,16,1,power_law_1.01,0.8036671638488769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,16,1,power_law_1.01,0.23040640354156494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,16,power_law_1.2,5.323923110961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,16,1,power_law_1.01,0.24160640239715575
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,16,power_law_1.2,0.29713919162750246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,16,1,power_law_1.2,0.03139199912548065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,16,1,power_law_1.01,0.2648767948150635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,16,1,power_law_1.2,0.03528960049152374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,16,power_law_1.2,0.41637120246887205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,16,1,power_law_1.01,0.2948096036911011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,16,1,power_law_1.2,0.07015680074691773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,16,1,power_law_1.01,0.31135358810424807
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,16,power_law_1.2,0.5470848083496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,16,1,power_law_1.2,0.09063040018081665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,16,1,power_law_1.01,0.34517760276794435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,16,1,power_law_1.2,0.11509759426116943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,16,power_law_1.2,0.8027008056640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,16,1,power_law_1.2,0.1540735960006714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,16,1,power_law_1.01,0.9577728271484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,16,1,power_law_1.2,0.20495998859405518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,16,1,power_law_1.2,0.034771201014518735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,16,1,power_law_1.2,0.21040639877319336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,16,1,power_law_1.2,0.04047360122203827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,16,1,power_law_1.01,1.290867233276367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,16,1,power_law_1.2,0.06683520078659058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,16,1,power_law_1.2,0.21660799980163575
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,16,1,power_law_1.01,0.028230398893356323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,16,1,power_law_1.2,0.0852288007736206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,16,1,power_law_1.2,0.22056961059570312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,16,1,power_law_1.01,0.029996800422668456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,16,1,power_law_1.01,1.6279424667358398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,16,1,power_law_1.2,0.1076416015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,16,1,power_law_1.01,0.38949759006500245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,16,1,power_law_1.2,0.2515199899673462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,16,1,power_law_1.01,0.03358719944953918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,16,1,power_law_1.2,0.13602559566497802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,16,1,power_law_1.2,0.26857600212097166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,16,1,power_law_1.01,0.4370175838470459
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,16,1,power_law_1.01,0.04307839870452881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,16,1,power_law_1.2,0.1768704056739807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,16,1,power_law_1.2,0.28044800758361815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,16,power_law_1.2,1.152672004699707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,16,1,power_law_1.01,0.05941759943962097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,16,1,power_law_1.2,0.18468480110168456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,16,1,power_law_1.2,0.30185599327087403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,16,1,power_law_1.01,0.07777919769287109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,16,1,power_law_1.2,0.2037951946258545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,16,1,power_law_1.2,0.3398207902908325
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,16,1,power_law_1.01,0.09093760251998902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,16,1,power_law_1.2,0.35669119358062745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,16,1,power_law_1.01,0.09757440090179444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,16,1,power_law_1.2,0.38103039264678956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,16,1,power_law_1.01,0.10076800584793091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,16,1,power_law_1.2,0.46306557655334474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,16,1,power_law_1.01,0.10485119819641113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,16,1,power_law_1.2,0.047302401065826415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,16,1,power_law_1.01,0.11018240451812744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,16,1,power_law_1.01,0.5092288017272949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,16,1,power_law_1.01,2.3036672592163088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,16,1,power_law_1.2,0.05221760272979736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,16,1,power_law_1.01,0.11598720550537109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,16,1,power_law_1.01,0.5866112232208252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,16,1,power_law_1.2,0.07299839854240417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,16,1,power_law_1.01,0.12491519451141357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,16,1,power_law_1.2,0.20917119979858398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,16,1,power_law_1.2,0.09735680222511292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,16,1,power_law_1.01,0.7653696060180664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,16,1,power_law_1.2,0.22561280727386473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,16,1,power_law_1.2,0.13362560272216797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,16,1,power_law_1.01,2.986214447021484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,16,1,power_law_1.2,0.23541760444641113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,16,1,balanced,0.04786666731039683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,16,1,power_law_1.01,0.9313216209411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,16,1,power_law_1.2,0.18221440315246581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,16,1,power_law_1.2,0.24682240486145018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,16,1,power_law_1.2,0.24929280281066896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,16,1,power_law_1.01,1.2607232093811036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,16,1,power_law_1.2,0.2582655906677246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,16,1,power_law_1.2,0.26128640174865725
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,16,1,power_law_1.01,0.13845759630203247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,16,1,power_law_1.01,1.5995391845703124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,16,1,power_law_1.2,0.26229760646820066
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,16,1,power_law_1.01,0.15015679597854614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,16,1,power_law_1.2,0.27119998931884765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,16,1,power_law_1.01,0.16037119626998902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,16,1,power_law_1.2,0.279750394821167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,16,1,power_law_1.01,0.188864004611969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,16,1,power_law_1.01,2.2754560470581056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,16,1,power_law_1.2,0.5000127792358399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,16,1,power_law_1.2,0.2847935914993286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,16,1,power_law_1.2,0.2926719903945923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,16,1,power_law_1.2,0.5593599796295166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,16,1,power_law_1.2,0.3041215896606445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,16,1,power_law_1.2,0.3155904054641724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,16,1,power_law_1.01,2.949126434326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,16,balanced,0.02735466758410136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,16,1,power_law_1.2,0.34057600498199464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,16,balanced,0.029365333418051403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,16,balanced,0.027429332335789997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,16,balanced,0.031514666974544525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,16,balanced,0.0611413319905599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,16,1,power_law_1.2,0.3810751914978027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,16,1,power_law_1.2,0.43909759521484376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,16,1,power_law_1.2,0.26657919883728026
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,16,1,power_law_1.2,0.026752001047134398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,16,1,power_law_1.2,0.3007040023803711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,16,1,power_law_1.2,0.5135231971740722
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,16,1,power_law_1.2,0.02849920094013214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,16,1,power_law_1.2,0.31383678913116453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,16,1,power_law_1.2,0.03304960131645203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,16,1,power_law_1.2,0.6050047874450684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,16,1,power_law_1.2,0.6328512191772461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,16,1,power_law_1.2,0.041433599591255185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,16,1,power_law_1.2,0.7721856117248536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,16,1,power_law_1.2,0.817632007598877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,16,balanced,0.06711466610431671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,16,balanced,0.06963199873765309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,16,balanced,0.0702400008837382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,16,balanced,0.07250133156776428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,16,1,power_law_1.2,0.9438655853271485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,16,balanced,0.07250133156776428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,16,1,power_law_1.2,0.9627264022827149
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,16,1,power_law_1.01,0.18523520231246948
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,16,1,power_law_1.01,0.21519999504089354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,16,1,power_law_1.2,1.296230411529541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,16,1,power_law_1.2,1.2943552017211915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,16,1,power_law_1.2,0.3472831964492798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,16,1,power_law_1.2,0.40784640312194825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,16,1,power_law_1.2,1.6475008010864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,16,1,power_law_1.2,1.635366439819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,16,1,power_law_1.2,0.056857597827911374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,16,1,balanced,0.055717334151268005
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,16,1,power_law_1.2,0.07567999958992004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,16,balanced,0.07656000057856242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,16,balanced,0.0804746647675832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,16,balanced,0.08265066643555959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,16,balanced,0.08982400099436443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,16,balanced,0.09286933143933614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,16,1,power_law_1.2,2.3194047927856447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,16,balanced,0.09897599617640178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,16,balanced,0.11262933413187663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,16,1,power_law_1.01,0.23129599094390868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,16,balanced,0.12980266412099203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,16,balanced,0.1453333298365275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,16,balanced,0.18764267365137735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,16,1,power_law_1.01,0.2757312059402466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,16,balanced,0.22930665810902914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,16,balanced,0.3153120080629985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,16,1,power_law_1.01,0.3398272037506104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,16,balanced,0.4029279947280884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,16,balanced,0.5887200037638346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,16,1,power_law_1.01,0.4203648090362549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,16,1,balanced,0.07296533385912578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,16,1,balanced,0.1000426709651947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,16,balanced,0.7571520010630289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,16,1,power_law_1.01,0.49351038932800295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,16,1,power_law_1.2,0.09040639996528625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,16,balanced,1.1048160394032795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,16,1,power_law_1.01,0.6237567901611328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,16,balanced,1.4545226097106934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,16,balanced,0.029440000653266907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,16,balanced,0.03232000023126602
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,16,1,power_law_1.01,0.8812095642089843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,16,1,power_law_1.2,0.44053120613098146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,16,1,power_law_1.2,2.990630340576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,16,balanced,0.0377866675456365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,16,balanced,0.040181333820025124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,16,balanced,0.040463998913764954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,16,balanced,0.044341335693995156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,16,balanced,0.049695998430252075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,16,1,power_law_1.2,2.3297088623046873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,16,balanced,0.051701332132021584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,16,balanced,0.052042668064435325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,16,balanced,0.05381333331267039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,16,balanced,0.05463466544946035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,16,1,power_law_1.2,0.0972544014453888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,16,balanced,0.05602666735649109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,16,balanced,0.05996266504128774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,16,balanced,0.062368000547091164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,16,balanced,0.06330666442712148
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,16,1,power_law_1.2,0.10055680274963379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,16,balanced,0.07050666709740956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,16,balanced,0.07637866834799449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,16,balanced,0.08497066299120586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,16,1,power_law_1.2,0.10558079481124878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,16,balanced,0.09470933675765991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,16,balanced,0.11486400167147319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,16,1,power_law_1.2,3.0229631423950196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,16,balanced,0.13875733812650046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,16,1,power_law_1.2,0.10991359949111938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,16,balanced,0.18195732434590658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,16,balanced,0.2247999906539917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,16,1,power_law_1.01,1.146662425994873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,16,1,power_law_1.2,0.5179711818695069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,16,power_law_1.01,0.032467201352119446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,16,1,power_law_1.2,0.6025023937225342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,16,power_law_1.01,0.035955199599266054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,16,power_law_1.01,0.03804160058498383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,16,1,power_law_1.2,0.7684288024902344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,16,power_law_1.01,0.03603839874267578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,16,balanced,0.03356266766786575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,16,balanced,0.03356799980004629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,16,balanced,0.0611413319905599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,16,1,power_law_1.2,0.9360320091247558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,16,balanced,0.06229333579540253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,16,balanced,0.06422399977842967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,16,balanced,0.0643039991458257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,16,balanced,0.064410666624705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,16,balanced,0.06899199883143108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,16,balanced,0.07045333087444305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,16,1,power_law_1.2,0.11793919801712036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,16,1,power_law_1.2,1.2700480461120605
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,16,1,power_law_1.2,0.12406400442123414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,16,1,power_law_1.2,0.1386944055557251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,16,1,power_law_1.2,1.6035520553588867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,16,1,power_law_1.2,0.15256960391998292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,16,1,power_law_1.2,0.1647871971130371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,16,power_law_1.01,0.05985280275344849
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,16,balanced,0.02536533276240031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,16,power_law_1.01,0.06050559878349304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,16,balanced,0.02534399926662445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,16,balanced,0.02531733363866806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,16,power_law_1.01,0.06384639739990235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,16,balanced,0.02736533433198929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,16,1,power_law_1.2,2.2800512313842773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,16,balanced,0.027658666173617046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,16,power_law_1.01,0.06536319851875305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,16,balanced,0.07366399963696797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,16,power_law_1.01,0.06846079826354981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,16,balanced,0.07791999975840251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,16,balanced,0.08274133503437042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,16,power_law_1.01,0.06844800114631652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,16,balanced,0.088837335507075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,16,balanced,0.09508267045021057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,16,power_law_1.01,0.07254400253295898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,16,power_law_1.01,0.07619839906692505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,16,1,power_law_1.2,2.9599231719970702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,16,power_law_1.01,0.07832959890365601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,16,power_law_1.01,0.08603519797325135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,16,power_law_1.01,0.032691198587417605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,16,1,balanced,0.1618613302707672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,16,power_law_1.01,0.0939136028289795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,16,1,balanced,0.29199467102686566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,16,power_law_1.01,0.09968640208244324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,16,balanced,0.029877332349618275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,16,balanced,0.033301333586374916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,16,balanced,0.3155146638552348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,16,balanced,0.03355200091997782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,16,balanced,0.035418666899204254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,16,balanced,0.40223467350006104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,16,balanced,0.5765866835912069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,16,balanced,0.10957866907119751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,16,balanced,0.12909332911173502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,16,balanced,0.7468000253041586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,16,balanced,1.0923093159993489
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,16,1,power_law_1.2,0.1900607943534851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,16,balanced,1.4383893013000488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,16,1,balanced,0.2929439942042033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,16,1,balanced,0.29548267523447674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,16,power_law_1.01,0.1115007996559143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,16,power_law_1.01,0.03696640133857727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,16,power_law_1.01,0.0416703999042511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,16,power_law_1.01,0.04374400079250336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,16,power_law_1.01,0.13077759742736816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,16,power_law_1.01,0.04371840059757233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,16,balanced,0.1529813309510549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,16,balanced,0.19731199741363525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,16,power_law_1.01,0.14871679544448851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,16,power_law_1.01,0.04738560020923614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,16,balanced,0.24288000663121542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,16,power_law_1.01,0.048742398619651794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,16,power_law_1.01,0.05059840083122254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,16,power_law_1.01,0.03407360017299652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,16,power_law_1.01,0.05146239995956421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,16,power_law_1.01,0.03495680093765259
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,16,1,power_law_1.2,0.1878335952758789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,16,power_law_1.01,0.05363199710845947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,16,1,balanced,0.2958986759185791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,16,power_law_1.01,0.03408640027046204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,16,1,balanced,0.2960746685663859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,16,power_law_1.01,0.0545087993144989
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,16,1,power_law_1.2,0.21976959705352783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,16,balanced,0.035760000348091125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,16,power_law_1.01,0.05857920050621033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,16,balanced,0.035461333890755974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,16,power_law_1.01,0.05997440218925476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,16,1,power_law_1.2,0.2326143980026245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,16,power_law_1.01,0.061401599645614625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,16,1,power_law_1.2,0.2787775993347168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,16,power_law_1.01,0.06306560039520263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,16,power_law_1.01,0.06401280164718628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,16,power_law_1.01,0.19570560455322267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,16,1,power_law_1.2,0.34565119743347167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,16,power_law_1.01,0.06586880087852479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,16,power_law_1.01,0.2540479898452759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,16,power_law_1.01,0.06783360242843628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,16,1,power_law_1.2,0.42691841125488283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,16,power_law_1.01,0.07126399874687195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,16,1,power_law_1.2,0.49886717796325686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,16,1,balanced,0.3035946687062581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,16,power_law_1.01,0.05767040252685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,16,1,balanced,0.3060426712036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,16,balanced,0.05605866511662801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,16,1,balanced,0.308186670144399
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,16,1,power_law_1.2,0.6278079986572266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,16,power_law_1.01,0.06124160289764404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,16,1,balanced,0.3163520097732544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,16,power_law_1.01,0.06263039708137512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,16,1,balanced,0.32241066296895343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,16,1,balanced,0.3304426670074463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,16,power_law_1.01,0.06968320012092591
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,16,1,power_law_1.2,0.8680768013000488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,16,power_law_1.01,0.07536640167236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,16,power_law_1.01,0.08352000117301941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,16,power_law_1.01,0.09820799827575684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,16,power_law_1.01,0.3493760108947754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,16,power_law_1.01,0.12460800409317016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,16,power_law_1.01,0.07519360184669495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,16,power_law_1.01,0.46185598373413084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,16,power_law_1.01,0.147161602973938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,16,power_law_1.01,0.07938560247421264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,16,power_law_1.01,0.19811840057373048
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,16,balanced,0.05612266560395559
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,16,balanced,0.04391466577847799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,16,power_law_1.01,0.24442241191864014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,16,power_law_1.01,0.3432447910308838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,16,1,balanced,0.3431359926859538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,16,1,balanced,0.3641973336537679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,16,power_law_1.01,0.43355522155761717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,16,1,balanced,0.38813332716623944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,16,1,balanced,0.43516798814137775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,16,1,power_law_1.2,1.1677568435668946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,16,power_law_1.01,0.6520512104034424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,16,1,balanced,0.4952640136082967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,16,1,balanced,0.7469866275787354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,16,power_law_1.01,0.8185791969299316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,16,1,balanced,0.8483893076578776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,16,power_law_1.01,0.08560640215873719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,16,power_law_1.01,0.6576767921447754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,16,power_law_1.01,0.09134079813957215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,16,balanced,0.03976533313592275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,16,balanced,0.03996799886226654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,16,1,balanced,1.2120693524678547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,16,power_law_1.01,0.10364799499511719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,16,power_law_1.01,1.2221440315246581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,16,power_law_1.01,0.842950439453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,16,balanced,0.3404159943262736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,16,power_law_1.01,0.13146239519119263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,16,balanced,0.43627198537190753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,16,1,balanced,1.5567626953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,16,power_law_1.01,0.1545024037361145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,16,balanced,0.6290080149968466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,16,power_law_1.01,1.2506879806518554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,16,balanced,0.8152639865875244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,16,1,balanced,2.2747626304626465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,16,balanced,1.188442627588908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,16,power_law_1.01,1.6631359100341796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,16,power_law_1.2,0.034431999921798705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,16,1,balanced,2.977856000264486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,16,balanced,0.041759997606277466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,16,power_law_1.2,0.036447998881340024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,16,balanced,0.04374399781227112
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,16,balanced,0.04428799947102865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,16,power_law_1.2,0.036697599291801455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,16,balanced,0.050213331977526345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,16,balanced,0.055589333176612854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,16,power_law_1.2,0.03570559918880463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,16,power_law_1.01,0.20609920024871825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,16,power_law_1.2,0.058246397972106935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,16,power_law_1.01,0.024639999866485594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,16,power_law_1.01,0.26528639793395997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,16,power_law_1.2,0.06058239936828613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,16,power_law_1.01,0.024294400215148927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,16,power_law_1.2,0.06561279892921448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,16,power_law_1.01,0.3714495897293091
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,16,power_law_1.01,0.02475520074367523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,16,power_law_1.2,0.0680512011051178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,16,power_law_1.01,0.47087998390197755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,16,power_law_1.01,0.02542720139026642
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,16,power_law_1.01,0.02593280076980591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,16,power_law_1.01,0.7077248096466064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,16,power_law_1.01,0.027827200293540955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,16,power_law_1.2,0.032742398977279666
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,16,power_law_1.01,0.030502399802207945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,16,balanced,0.0718560020128886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,16,power_law_1.01,0.9006464004516601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,16,power_law_1.01,1.6251903533935548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,16,power_law_1.2,0.03654400110244751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,16,power_law_1.01,1.3606975555419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,16,power_law_1.2,0.04134399890899658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,16,power_law_1.2,0.044147199392318724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,16,power_law_1.2,0.06900479793548583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,16,balanced,1.5640746752421062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,16,power_law_1.2,0.0435263991355896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,16,power_law_1.2,0.0707584023475647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,16,power_law_1.01,1.750739288330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,16,power_law_1.2,0.07314559817314148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,16,power_law_1.2,0.07736319899559022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,16,power_law_1.2,0.03420799970626831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,16,balanced,0.08504533767700195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,16,balanced,0.12090133627255757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,16,power_law_1.2,0.03484799861907959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,16,balanced,0.15332266688346863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,16,balanced,0.21199999252955118
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,16,balanced,0.27928000688552856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,16,balanced,0.0420959989229838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,16,balanced,0.04162666698296865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,16,balanced,0.04167466859022776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,16,balanced,0.06016000111897787
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,16,power_law_1.2,0.025235199928283693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,16,power_law_1.2,0.024864000082015992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,16,balanced,0.0360959991812706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,16,balanced,0.03572800010442734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,16,power_law_1.2,0.025260800123214723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,16,balanced,0.043882668018341064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,16,balanced,0.06814399858315785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,16,balanced,0.13184000054995218
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,16,power_law_1.2,0.02654080092906952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,16,power_law_1.2,0.08101760149002075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,16,power_law_1.01,0.03139199912548065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,16,power_law_1.2,0.0889792025089264
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,16,power_law_1.01,0.03253119885921478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,16,power_law_1.2,0.09388160109519958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,16,power_law_1.01,0.032492798566818235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,16,power_law_1.2,0.1011072039604187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,16,power_law_1.01,0.03289600014686585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,16,power_law_1.2,0.11096960306167603
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,16,power_law_1.01,0.04975999891757965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,16,balanced,0.11497599879900615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,16,balanced,0.11681600411732991
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,16,power_law_1.01,0.04673919975757599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,16,power_law_1.2,0.048153600096702574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,16,power_law_1.01,0.03808639943599701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,16,power_law_1.2,0.04832000136375427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,16,power_law_1.2,0.05103359818458557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,16,balanced,0.1327839990456899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,16,balanced,0.13723733027776083
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,16,power_law_1.01,0.04301440119743347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,16,balanced,0.138373335202535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,16,balanced,0.14008532961209616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,16,balanced,0.1426400045553843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,16,power_law_1.2,0.033542400598526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,16,power_law_1.01,0.03788160085678101
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,16,power_law_1.01,0.03945600092411041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,16,power_law_1.01,0.04306559860706329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,16,power_law_1.2,0.13118720054626465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,16,power_law_1.01,0.047328001260757445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,16,balanced,0.11952533324559529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,16,power_law_1.2,0.1530751943588257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,16,balanced,0.12128532926241557
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,16,power_law_1.01,0.055251199007034305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,16,power_law_1.2,0.2070528030395508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,16,power_law_1.01,0.06475520133972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,16,power_law_1.01,0.0821120023727417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,16,power_law_1.2,0.026099199056625368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,16,power_law_1.2,0.25559039115905763
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,16,power_law_1.01,0.10410879850387574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,16,power_law_1.2,0.0286080002784729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,16,power_law_1.2,0.3734656095504761
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,16,power_law_1.2,0.031007999181747438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,16,power_law_1.2,0.058233600854873654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,16,power_law_1.2,0.03216640055179596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,16,power_law_1.2,0.060159999132156375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,16,power_law_1.2,0.03318400084972382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,16,power_law_1.2,0.03338240087032318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,16,balanced,0.12462932864824931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,16,balanced,0.12611200412114462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,16,balanced,0.13199999928474426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,16,balanced,0.13572800159454346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,16,balanced,0.14196266730626425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,16,power_law_1.2,0.051481598615646364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,16,power_law_1.2,0.053651201725006106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,16,balanced,0.14792533715566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,16,power_law_1.2,0.05432320237159729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,16,balanced,0.1524799962838491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,16,balanced,0.15851733088493347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,16,power_law_1.2,0.0575872004032135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,16,power_law_1.2,0.061913597583770755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,16,power_law_1.2,0.06218879818916321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,16,power_law_1.2,0.06296319961547851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,16,power_law_1.2,0.06335359811782837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,16,power_law_1.2,0.06962559819221496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,16,power_law_1.2,0.0645632028579712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,16,power_law_1.2,0.07740160226821899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,16,power_law_1.2,0.06628479957580566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,16,power_law_1.2,0.08640000224113464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,16,balanced,0.15017066399256387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,16,power_law_1.2,0.06881279945373535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,16,balanced,0.16057067116101584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,16,power_law_1.2,0.09900799989700318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,16,balanced,0.17150400082270303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,16,power_law_1.2,0.072326397895813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,16,power_law_1.2,0.12526079416275024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,16,power_law_1.01,0.14012160301208496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,16,power_law_1.2,0.07573760151863099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,16,power_law_1.2,0.1547327995300293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,16,balanced,0.16631999611854553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,16,power_law_1.2,0.08122239708900451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,16,balanced,0.17541333039601645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,16,power_law_1.01,0.1835904002189636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,16,balanced,0.18578133980433145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,16,power_law_1.2,0.08699520230293274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,16,power_law_1.2,0.1991871953010559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,16,balanced,0.20615466435750326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,16,balanced,0.24131733179092407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,16,power_law_1.01,0.2517375946044922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,16,power_law_1.2,0.24988799095153807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,16,balanced,0.2792853315671285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,16,balanced,0.3893333276112874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,16,power_law_1.2,0.03283840119838714
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,16,power_law_1.01,0.33619840145111085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,16,balanced,0.47862398624420166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,16,balanced,0.05195199946562449
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,16,power_law_1.2,0.05006719827651977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,16,balanced,0.05525333185990652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,16,balanced,0.6977120240529379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,16,balanced,0.06192000210285187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,16,balanced,0.06835199892520905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,16,power_law_1.2,0.04565759897232056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,16,balanced,0.0902933379014333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,16,balanced,0.8803573449452718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,16,balanced,0.191210667292277
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,16,power_law_1.2,0.0395904004573822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,16,balanced,0.23745065927505493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,16,power_law_1.2,0.3531519889831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,16,balanced,0.2812906702359517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,16,balanced,1.2794773578643799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,16,balanced,0.3988746802012126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,16,balanced,0.49127999941507977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,16,balanced,1.6640960375467937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,16,power_law_1.2,0.09434239864349366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,16,balanced,0.7016800244649252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,16,power_law_1.2,0.10814720392227173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,16,balanced,0.8992106914520264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,16,power_law_1.2,0.13253120183944703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,16,balanced,2.513376077016195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,16,power_law_1.2,0.1584831953048706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,16,balanced,1.3076106707255046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,16,power_law_1.2,0.21590399742126465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,16,balanced,0.09102400143941243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,16,balanced,0.09356266260147095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,16,power_law_1.2,0.4579455852508545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,16,balanced,1.709536075592041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,16,power_law_1.2,0.28083200454711915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,16,power_law_1.2,0.45020160675048826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,16,balanced,3.389535903930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,16,power_law_1.2,0.3823168039321899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,16,balanced,2.5254613558451333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,16,power_law_1.2,0.5023551940917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,16,power_law_1.2,0.7071167945861816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,16,balanced,3.3340587615966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,16,balanced,0.09622933467229207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,16,power_law_1.01,0.08202239871025085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,16,power_law_1.2,0.03927040100097656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,16,power_law_1.01,0.08186240196228027
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,16,power_law_1.2,0.03914879858493805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,16,power_law_1.01,0.08661760091781616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,16,power_law_1.2,0.04000000059604645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,16,power_law_1.01,0.07951359748840332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,16,power_law_1.2,0.04694400131702423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,16,power_law_1.01,0.1290176033973694
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,16,power_law_1.2,0.049804800748825075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,16,power_law_1.01,0.1307136058807373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,16,power_law_1.2,0.05783039927482605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,16,power_law_1.01,0.13501440286636351
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,16,power_law_1.2,0.0657151997089386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,16,power_law_1.01,0.13649280071258546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,16,power_law_1.2,0.08703359961509705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,16,power_law_1.01,0.14041600227355958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,16,balanced,0.09915733337402344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,16,balanced,0.10236799716949463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,16,power_law_1.01,0.14139519929885863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,16,power_law_1.2,0.10961920022964478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,16,balanced,0.10566932956377666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,16,balanced,0.11347732941309611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,16,power_law_1.01,0.14543360471725464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,16,balanced,0.11798399686813354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,16,balanced,0.13009066383043924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,16,power_law_1.01,0.15155839920043945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,16,power_law_1.01,0.15693440437316894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,16,power_law_1.01,0.1673151969909668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,16,power_law_1.01,0.17845760583877562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,16,balanced,0.027493332823117573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,16,power_law_1.01,0.19048960208892823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,16,power_law_1.01,0.21200640201568605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,16,power_law_1.01,0.26883199214935305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,16,power_law_1.01,0.3432832002639771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,16,balanced,0.14040533701578775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,16,balanced,0.15366400281588236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,16,power_law_1.01,0.46108160018920896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,16,balanced,0.17522666851679483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,16,balanced,0.22002132733662924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,16,balanced,0.267797331015269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,16,balanced,0.37402665615081787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,16,balanced,0.4597173531850179
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,16,balanced,0.027530667682488758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,16,balanced,0.028197333216667175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,16,balanced,0.02923733244339625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,16,balanced,0.6485226551691691
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,16,balanced,0.03378133227427801
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,16,balanced,0.036101333796978
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,16,balanced,0.039493332306543984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,16,balanced,0.8336373170216879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,16,balanced,1.2107840379079182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,16,power_law_1.2,0.15166720151901245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,16,balanced,1.5861652692159016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,16,power_law_1.01,0.5931136131286621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,16,power_law_1.2,0.6880512237548828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,16,power_law_1.01,0.8431679725646972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,16,balanced,2.3281280199686685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,16,1,power_law_1.01,0.04666880071163178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,16,balanced,0.03987200061480204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,16,power_law_1.01,1.0856575965881348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,16,balanced,0.06158400078614553
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,16,balanced,0.06058133145173391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,16,balanced,0.05203733344872793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,16,balanced,3.079178810119629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,16,power_law_1.01,1.5454976081848144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,16,power_law_1.01,0.06925439834594727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,16,power_law_1.01,0.0751039981842041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,16,power_law_1.01,0.0745024025440216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,16,power_law_1.01,0.07489920258522034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,16,power_law_1.01,2.194175910949707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,16,power_law_1.01,0.11070719957351685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,16,power_law_1.01,0.11481599807739258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,16,power_law_1.01,0.11796480417251587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,16,balanced,0.12310399611790974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,16,power_law_1.01,0.11987839937210083
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,16,power_law_1.2,0.19667840003967285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,16,power_law_1.01,0.12167680263519287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,16,power_law_1.01,3.102943992614746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,16,power_law_1.01,0.12469120025634765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,16,power_law_1.01,0.12972160577774047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,16,power_law_1.2,0.6704832077026367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,16,power_law_1.2,0.9534912109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,16,power_law_1.01,4.210713577270508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,16,balanced,0.10658666491508484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,16,balanced,0.08242666721343994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,16,power_law_1.01,0.06812160015106201
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,16,balanced,0.06580266853173573
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,16,balanced,0.06690666576226552
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,16,balanced,0.06268799801667531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,16,power_law_1.01,0.06974080204963684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,16,balanced,0.07022933165232341
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,16,balanced,0.07461333274841309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,16,power_law_1.01,0.07645440101623535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,16,balanced,0.08673066894213359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,16,balanced,0.09955199559529622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,16,power_law_1.2,0.8548352241516113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,16,power_law_1.01,0.13542399406433106
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,16,balanced,0.12967466314633688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,16,balanced,0.15451199809710184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,16,power_law_1.01,0.14160000085830687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,16,balanced,0.21186665693918863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,16,balanced,0.2752426664034526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,16,balanced,0.3914506832758586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,16,balanced,0.5080053408940634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,16,power_law_1.2,0.9176959991455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,16,power_law_1.01,0.07619199752807618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,16,power_law_1.01,0.08451200127601624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,16,power_law_1.01,0.15802240371704102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,16,power_law_1.01,0.08818560242652893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,16,power_law_1.01,0.17677439451217652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,16,1,power_law_1.01,0.05238400101661682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,16,power_law_1.01,0.1869503974914551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,16,1,power_law_1.01,0.0726207971572876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,16,power_law_1.01,0.2116096019744873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,16,1,power_law_1.01,0.09894400238990783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,16,power_law_1.01,0.2736063957214355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,16,power_law_1.01,0.33703041076660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,16,power_law_1.01,0.45797119140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,16,power_law_1.01,0.03285120129585266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,16,power_law_1.01,0.5874559879302979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,16,power_law_1.01,0.09011840224266052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,16,power_law_1.01,0.03153280019760132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,16,power_law_1.01,0.09134719967842102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,16,power_law_1.01,0.03155840039253235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,16,power_law_1.01,0.8325823783874512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,16,power_law_1.01,0.09768319725990296
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,16,power_law_1.01,0.032441601157188416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,16,power_law_1.01,0.09838079810142517
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,16,power_law_1.01,0.03225600123405457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,16,power_law_1.01,1.0791744232177733
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,16,power_law_1.01,0.03491199910640717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,16,power_law_1.01,0.03738240003585815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,16,power_law_1.01,0.03930239975452423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,16,power_law_1.01,1.599513626098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,16,power_law_1.01,0.055340802669525145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,16,power_law_1.01,0.05361279845237732
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,16,power_law_1.01,0.047353601455688475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,16,power_law_1.01,0.10213760137557984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,16,power_law_1.01,2.1097280502319338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,16,power_law_1.01,0.08353919982910156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,16,power_law_1.01,0.10408320426940917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,16,power_law_1.01,0.07542399764060974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,16,power_law_1.01,0.06305279731750488
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,16,power_law_1.01,0.06223359704017639
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,16,power_law_1.01,0.06308479905128479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,16,power_law_1.01,3.0918336868286134
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,16,power_law_1.01,0.0720192015171051
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,16,power_law_1.01,0.07820799946784973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,16,power_law_1.01,0.0953279972076416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,16,power_law_1.01,0.11061760187149047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,16,power_law_1.01,0.14529919624328613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,16,power_law_1.01,0.1090432047843933
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,16,power_law_1.01,0.1803455948829651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,16,power_law_1.01,4.182169723510742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,16,power_law_1.01,0.11793919801712036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,16,power_law_1.01,0.2523519992828369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,16,power_law_1.2,0.07861120104789734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,16,power_law_1.01,0.13377280235290528
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,16,power_law_1.01,0.32572159767150877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,16,power_law_1.01,0.14665600061416625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,16,power_law_1.01,0.15870720148086548
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,16,power_law_1.01,0.47435522079467773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,16,power_law_1.01,0.1870527982711792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,16,power_law_1.01,0.24523520469665527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,16,power_law_1.01,0.6478591918945312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,16,power_law_1.01,0.2951551914215088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,16,power_law_1.2,0.06832000017166137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,16,power_law_1.2,0.0695360004901886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,16,power_law_1.01,0.4176447868347168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,16,power_law_1.2,0.07453439831733703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,16,power_law_1.01,0.5205952167510987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,16,power_law_1.2,0.07108479738235474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,16,power_law_1.2,0.07902079820632935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,16,power_law_1.2,0.10920959711074829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,16,power_law_1.2,0.08464639782905578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,16,power_law_1.01,0.7391615867614746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,16,power_law_1.2,0.11370879411697388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,16,power_law_1.2,0.07555840015411378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,16,power_law_1.2,0.11678080558776856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,16,power_law_1.2,0.1226304054260254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,16,power_law_1.01,0.9667455673217773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,16,power_law_1.2,0.11871999502182007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,16,power_law_1.2,0.13070720434188843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,16,power_law_1.2,0.1211583971977234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,16,power_law_1.2,0.13345919847488402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,16,power_law_1.2,0.12275840044021606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,16,power_law_1.2,0.1362496018409729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,16,power_law_1.01,1.4014399528503418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,16,power_law_1.2,0.12840960025787354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,16,power_law_1.2,0.1371135950088501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,16,power_law_1.2,0.13432960510253905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,16,power_law_1.2,0.14087040424346925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,16,power_law_1.2,0.14941439628601075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,16,power_law_1.2,0.1439743995666504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,16,power_law_1.2,1.3378239631652833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,16,power_law_1.01,1.8764863967895509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,16,power_law_1.2,0.16220799684524537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,16,power_law_1.2,0.1503551959991455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,16,power_law_1.2,0.1727679967880249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,16,power_law_1.2,0.1614400029182434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,16,power_law_1.2,0.19015040397644042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,16,power_law_1.2,0.16843520402908324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,16,power_law_1.2,0.21342079639434813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,16,power_law_1.2,0.17860480546951293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,16,power_law_1.2,0.2944191932678223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,16,power_law_1.01,2.700051116943359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,16,1,power_law_1.01,0.1409216046333313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,16,power_law_1.2,0.36741759777069094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,16,1,power_law_1.01,0.19114880561828612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,16,1,power_law_1.01,0.25036799907684326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,16,1,power_law_1.01,0.25884160995483396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,16,power_law_1.01,3.669516754150391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,16,1,power_law_1.01,0.26138880252838137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,16,power_law_1.2,0.06779519915580749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,16,1,power_law_1.01,0.264192008972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,16,power_law_1.2,0.06792320013046264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,16,1,power_law_1.01,0.2728192090988159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,16,power_law_1.2,0.07479680180549622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,16,power_law_1.2,0.49813117980957033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,16,1,power_law_1.01,0.279200005531311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,16,power_law_1.2,0.6368127822875976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,16,power_law_1.2,0.8509311676025391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,16,power_law_1.2,0.19203840494155883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,16,power_law_1.2,0.21617279052734376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,16,power_law_1.2,0.31182079315185546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,16,power_law_1.2,1.151033592224121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,16,power_law_1.2,0.36408960819244385
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,16,power_law_1.2,0.27026560306549074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,16,1,power_law_1.01,0.2850624084472656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,16,power_law_1.2,0.4714240074157715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,16,1,power_law_1.01,0.29407999515533445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,16,power_law_1.2,0.632755184173584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,16,1,power_law_1.01,0.3037823915481567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,16,power_law_1.2,0.8486720085144043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,16,power_law_1.2,0.07336959838867188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,16,power_law_1.2,0.07920640110969543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,16,power_law_1.2,0.08711040019989014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,16,power_law_1.2,1.316716766357422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,16,power_law_1.2,0.09081599712371827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,16,power_law_1.2,1.2170559883117675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,16,power_law_1.2,0.09263359904289245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,16,power_law_1.2,0.09696000218391418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,16,1,power_law_1.01,0.3133888006210327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,16,power_law_1.2,0.09795839786529541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,16,power_law_1.2,1.7788095474243164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,16,1,power_law_1.01,0.3321536064147949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,16,power_law_1.2,0.10510079860687256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,16,1,power_law_1.01,0.36778879165649414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,16,1,power_law_1.01,0.4303487777709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,16,1,power_law_1.01,0.5114687919616699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,16,1,power_law_1.01,0.5825600147247314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,16,power_law_1.2,0.1103551983833313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,16,1,power_law_1.01,0.7568704128265381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,16,power_law_1.2,0.11845120191574096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,16,power_law_1.2,1.7388799667358399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,16,power_law_1.2,0.13294719457626342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,16,1,power_law_1.01,0.9332159996032715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,16,power_law_1.2,0.14885120391845702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,16,power_law_1.2,0.16421760320663453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,16,power_law_1.2,0.19560960531234742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,16,power_law_1.2,0.25000319480895994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,16,power_law_1.2,0.3091007947921753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,16,power_law_1.2,0.42777600288391116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,16,power_law_1.2,0.5536064147949219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,16,1,power_law_1.01,1.2851584434509278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,16,power_law_1.2,0.7730432033538819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,16,1,power_law_1.01,1.631648063659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,16,power_law_1.2,1.0040448188781739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,16,1,power_law_1.01,2.3198783874511717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,16,1,power_law_1.01,3.016223907470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,16,power_law_1.2,2.287603187561035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,16,power_law_1.2,1.5209600448608398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,16,power_law_1.2,0.03155199885368347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,16,power_law_1.2,0.03207040131092072
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,16,power_law_1.2,0.031891199946403506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,16,power_law_1.2,0.030112001299858093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,16,power_law_1.2,0.03188480138778686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,16,power_law_1.2,0.03480319976806641
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,16,power_law_1.2,0.038700801134109494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,16,power_law_1.2,0.040678399801254275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,16,power_law_1.2,0.054764801263809205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,16,power_law_1.2,0.054604798555374146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,16,power_law_1.2,0.04792320132255554
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,16,power_law_1.2,0.08829439878463745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,16,power_law_1.2,0.08316159844398499
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,16,power_law_1.2,0.05854719877243042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,16,power_law_1.2,0.06217600107192993
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,16,power_law_1.2,0.06871680021286011
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,16,power_law_1.2,0.06280959844589233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,16,power_law_1.2,0.07025279998779296
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,16,power_law_1.2,0.08182399868965148
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,16,power_law_1.2,0.09859840273857116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,16,power_law_1.2,2.3281280517578127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,16,power_law_1.2,0.11910400390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,16,power_law_1.2,0.15454720258712767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,16,power_law_1.2,1.2479552268981933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,16,power_law_1.2,2.1268287658691407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,16,power_law_1.2,0.20357120037078857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,16,power_law_1.2,1.7131135940551758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,16,1,balanced,0.02720000098148982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,16,1,balanced,0.02996266633272171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,16,1,balanced,0.04761599997679392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,16,1,balanced,0.07044266661008199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,16,1,balanced,0.09851732850074768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,16,1,balanced,0.09941866993904114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,16,1,balanced,0.09915199875831604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,16,1,balanced,0.10124799609184265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,16,1,balanced,0.10322133700052898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,16,1,balanced,0.10181333621342976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,16,1,balanced,0.10532800356547038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,16,1,balanced,0.1442346672217051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,16,1,balanced,0.14896000425020853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,16,1,balanced,0.1520799994468689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,16,1,balanced,0.15546666582425436
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,16,power_law_1.2,0.29015679359436036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,16,1,balanced,0.1614400049050649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,16,1,balanced,0.17082667350769043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,16,1,balanced,0.1896160046259562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,16,1,balanced,0.21000534296035767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,16,1,balanced,0.25616000096003216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,16,1,balanced,0.2938773234685262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,16,1,balanced,0.4060639937718709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,16,1,balanced,0.5146986643473307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,16,1,balanced,0.7232053279876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,16,1,balanced,0.9272053241729736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,16,1,balanced,1.353866736094157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,16,1,balanced,1.7772639592488606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,16,1,balanced,0.03146133323510488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,16,1,balanced,0.03495466709136963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,16,1,balanced,0.039818666875362396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,16,1,balanced,0.06403733293215434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,16,1,balanced,0.09291199843088786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,16,1,balanced,0.09469866752624512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,16,1,balanced,0.09351467092831929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,16,1,balanced,0.0952959954738617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,16,1,balanced,0.09687999884287517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,16,1,balanced,0.0965173343817393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,16,1,balanced,0.09717333316802979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,16,1,balanced,0.1295093297958374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,16,1,balanced,0.1322879989941915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,16,1,balanced,0.13783466815948486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,16,1,balanced,0.14218133687973022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,16,1,balanced,0.14662399888038635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,16,1,balanced,0.1583199997742971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,16,1,balanced,0.17883733908335367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,16,1,balanced,0.19886932770411173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,16,1,balanced,0.26627200841903687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,16,1,balanced,0.30856533845265705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,16,1,balanced,0.4179840087890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,16,1,balanced,0.5280853509902954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,16,1,balanced,0.7461067040761312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,16,1,balanced,0.9627359708150228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,16,1,balanced,1.4114880561828613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,16,1,balanced,1.8698560396830242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,16,1,power_law_1.01,0.024908800423145295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,16,1,power_law_1.01,0.04076800048351288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,16,power_law_1.2,0.3651711940765381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,16,1,power_law_1.01,0.049472001194953916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,16,1,power_law_1.01,0.06307839751243591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,16,1,power_law_1.01,0.07276160120964051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,16,1,power_law_1.01,0.08923519849777221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,16,1,power_law_1.01,0.09099519848823548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,16,1,power_law_1.01,0.09404799938201905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,16,1,power_law_1.01,0.09816960096359253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,16,1,power_law_1.01,0.10043519735336304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,16,1,power_law_1.01,0.10734080076217652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,16,1,power_law_1.01,0.1412992000579834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,16,1,power_law_1.01,0.14512640237808228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,16,1,power_law_1.01,0.149619197845459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,16,1,power_law_1.01,0.15551359653472902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,16,1,power_law_1.01,0.16094080209732056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,16,1,power_law_1.01,0.17297919988632202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,16,power_law_1.2,3.3900543212890626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,16,1,power_law_1.01,0.1941375970840454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,16,1,power_law_1.01,0.21665918827056885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,16,1,power_law_1.01,0.2592384099960327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,16,1,power_law_1.01,0.31617920398712157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,16,1,power_law_1.01,0.4333824157714844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,16,1,power_law_1.01,0.5295936107635498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,16,1,power_law_1.01,0.7460608005523681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,16,1,power_law_1.01,0.9519295692443848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,16,1,power_law_1.01,1.3774784088134766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,16,1,power_law_1.01,1.803878402709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,16,1,power_law_1.01,0.029657599329948426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,16,1,power_law_1.01,0.03371520042419433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,16,1,power_law_1.01,0.04981119930744171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,16,1,power_law_1.01,0.06007680296897888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,16,1,power_law_1.01,0.06890239715576171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,16,1,power_law_1.01,0.084307199716568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,16,1,power_law_1.01,0.08822399973869324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,16,1,power_law_1.01,0.08839679956436157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,16,1,power_law_1.01,0.09244160056114196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,16,1,power_law_1.01,0.09500799775123596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,16,1,power_law_1.01,0.10106240510940552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,16,1,power_law_1.01,0.12628480195999145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,16,1,power_law_1.01,0.1306496024131775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,16,1,power_law_1.01,0.13543039560317993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,16,1,power_law_1.01,0.1415168046951294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,16,1,power_law_1.01,0.1473471999168396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,16,1,power_law_1.01,0.16219520568847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,16,1,power_law_1.01,0.18504960536956788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,16,1,power_law_1.01,0.21532158851623534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,16,1,power_law_1.01,0.27102720737457275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,16,1,power_law_1.01,0.333516788482666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,16,1,power_law_1.01,0.43996801376342776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,16,1,power_law_1.01,0.5547264099121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,16,1,power_law_1.01,0.774835205078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,16,1,power_law_1.01,0.999500846862793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,16,power_law_1.2,0.35255680084228513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,16,1,power_law_1.2,0.02476159930229187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,16,1,power_law_1.01,1.4473471641540527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,16,1,power_law_1.2,0.04354560077190399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,16,1,power_law_1.2,0.04922879934310913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,16,1,power_law_1.2,0.05959039926528931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,16,1,power_law_1.01,1.9034624099731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,16,1,power_law_1.2,0.06952959895133973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,16,1,power_law_1.2,0.029452800750732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,16,1,power_law_1.2,0.04165120124816894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,16,1,power_law_1.2,0.04927360117435455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,16,1,power_law_1.2,0.057715201377868654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,16,1,power_law_1.2,0.06589440107345582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,16,1,power_law_1.2,0.08432000279426574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,16,1,power_law_1.2,0.0868287980556488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,16,1,power_law_1.2,0.09081599712371827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,16,power_law_1.2,3.407148742675781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,16,1,power_law_1.2,0.09287040233612061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,16,1,power_law_1.2,0.08860160112380981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,16,1,power_law_1.2,0.09538559913635254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,16,1,power_law_1.2,0.09264000058174134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,16,1,power_law_1.2,0.10167039632797241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,16,1,power_law_1.2,0.09540479779243469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,16,1,power_law_1.2,0.1269376039505005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,16,1,power_law_1.2,0.09989759922027588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,16,1,power_law_1.2,0.13108479976654053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,16,1,power_law_1.2,0.10306559801101685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,16,1,power_law_1.2,0.1355455994606018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,16,1,power_law_1.2,0.10803840160369874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,16,1,power_law_1.2,0.1416640043258667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,16,1,power_law_1.2,0.14234880208969117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,16,1,power_law_1.2,0.148198401927948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,16,1,power_law_1.2,0.14809600114822388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,16,1,power_law_1.2,0.16289279460906983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,16,1,power_law_1.2,0.15107840299606323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,16,1,power_law_1.2,0.18794239759445192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,16,1,power_law_1.2,0.1567296028137207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,16,1,power_law_1.2,0.21614720821380615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,16,1,power_law_1.2,0.16227200031280517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,16,1,power_law_1.2,0.27100799083709715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,16,1,power_law_1.2,0.17429759502410888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,16,1,power_law_1.2,0.3369407892227173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,16,1,power_law_1.2,0.1945855975151062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,16,1,power_law_1.2,0.2176896095275879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,16,1,power_law_1.2,0.266758394241333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,16,1,power_law_1.2,0.3218816041946411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,16,1,power_law_1.2,0.43576960563659667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,16,1,power_law_1.2,0.44182400703430175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,16,1,power_law_1.2,0.5557439804077149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,16,1,power_law_1.2,0.7777728080749512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,16,1,power_law_1.2,0.5325439929962158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,16,1,power_law_1.2,1.0048768043518066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,16,1,power_law_1.2,0.7501311779022217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,16,power_law_1.2,3.021433639526367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,16,1,power_law_1.2,0.9607232093811036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,16,power_law_1.2,4.619059371948242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,16,balanced,0.0737013320128123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,16,balanced,0.0745600014925003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,16,balanced,0.08780266841252644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,16,balanced,0.1072266697883606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,16,balanced,0.206661323706309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,16,balanced,0.2712799906730652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,16,balanced,0.26236265897750854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,16,balanced,0.26586665709813434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,16,balanced,0.2608213424682617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,16,balanced,0.26160534222920734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,16,balanced,0.28074665864308673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,16,balanced,0.2970133423805237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,16,balanced,0.28134934107462567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,16,balanced,0.2961440086364746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,16,balanced,0.321450670560201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,16,balanced,0.3383520046869914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,16,balanced,0.36585601170857746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,16,balanced,0.4381706714630127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,16,balanced,0.4912319978078206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,16,balanced,0.7065280278523763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,16,balanced,0.8267680009206136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,16,balanced,1.1797866821289062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,16,balanced,1.496880054473877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,16,balanced,2.108821392059326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,16,1,power_law_1.2,1.4531392097473144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,16,power_law_1.2,1.8896448135375976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,16,balanced,2.9065332412719727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,16,balanced,0.049839998284975685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,16,balanced,0.0499839981396993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,16,balanced,0.075162669022878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,16,balanced,0.09247466921806335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,16,balanced,0.17036799589792886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,16,balanced,0.22470933198928833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,16,balanced,0.22453333934148154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,16,balanced,0.24515734116236368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,16,balanced,0.23241066932678223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,16,balanced,0.22822399934132895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,16,balanced,4.378975868225098
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,16,power_law_1.2,0.5493887901306153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,16,balanced,5.911946614583333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,16,balanced,0.06911466519037883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,16,balanced,0.07283733288447063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,16,balanced,0.09135466814041138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,16,balanced,0.09910399715105693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,16,balanced,0.15709333618481955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,16,balanced,0.1810986598332723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,16,balanced,0.18715733289718628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,16,balanced,0.18796267112096152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,16,1,power_law_1.2,1.3852224349975586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,16,balanced,0.19425066312154135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,16,balanced,0.19762667020161948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,16,balanced,0.20553600788116455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,16,balanced,0.21576533714930216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,16,balanced,0.22301334142684937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,16,balanced,0.24052266279856363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,16,balanced,0.25677333275477093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,16,balanced,0.27317333221435547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,16,balanced,0.3049973249435425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,16,balanced,0.24611733357111612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,16,balanced,0.36658668518066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,16,balanced,0.2486400008201599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,16,balanced,0.426581343015035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,16,balanced,0.2667306661605835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,16,balanced,0.282368004322052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,16,balanced,0.6086506843566895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,16,balanced,0.28481600681940716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,16,balanced,0.725536028544108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,16,balanced,1.0287360350290935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,16,balanced,1.3246080080668132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,16,balanced,1.865781307220459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,16,balanced,0.3035093347231547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,16,balanced,2.4596266746520996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,16,balanced,0.33450134595235187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,16,balanced,0.4037333329518636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,16,balanced,0.4770826498667399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,16,balanced,0.68177596728007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,16,balanced,3.588949203491211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,16,balanced,0.8132639726003011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,16,balanced,1.1515413125356038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,16,balanced,1.4863093694051106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,16,balanced,4.724831899007161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,16,power_law_1.2,4.5797119140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,16,balanced,2.0893173217773438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,16,power_law_1.01,0.14922879934310912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,16,power_law_1.01,0.19431040287017823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,16,balanced,2.758319854736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,16,power_law_1.01,0.1747648000717163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,16,power_law_1.01,0.19481600522994996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,16,power_law_1.01,0.19390720129013062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,16,power_law_1.01,0.22350080013275148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,16,balanced,4.034421284993489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,16,power_law_1.01,0.2407680034637451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,16,power_law_1.01,0.24564480781555176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,16,power_law_1.01,0.24374399185180665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,16,power_law_1.01,0.24972798824310302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,16,power_law_1.01,0.281004810333252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,16,power_law_1.01,0.2983423948287964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,16,power_law_1.01,0.2768768072128296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,16,power_law_1.01,0.3189183950424194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,16,power_law_1.01,0.3439232110977173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,16,power_law_1.01,0.417958402633667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,16,power_law_1.01,0.4516608238220215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,16,balanced,0.02956266701221466
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,16,balanced,0.029487999776999157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,16,power_law_1.01,0.5196159839630127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,16,balanced,5.536629358927409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,16,power_law_1.01,0.6084159851074219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,16,power_law_1.01,0.12480000257492066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,16,power_law_1.01,0.8101247787475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,16,power_law_1.01,0.15800319910049437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,16,power_law_1.01,0.14588160514831544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,16,power_law_1.01,1.0662336349487305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,16,power_law_1.01,0.15564160346984862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,16,power_law_1.01,0.17847039699554443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,16,balanced,0.03125333289305369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,16,power_law_1.01,1.4226367950439454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,16,power_law_1.01,0.22187519073486328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,16,power_law_1.01,0.21131520271301268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,16,power_law_1.01,0.22403841018676757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,16,power_law_1.01,1.923628807067871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,16,power_law_1.01,0.21911680698394775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,16,power_law_1.01,0.23004798889160155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,16,power_law_1.01,0.2328320026397705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,16,power_law_1.01,0.24897279739379882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,16,power_law_1.01,2.702060890197754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,16,power_law_1.01,0.2612799882888794
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,16,balanced,0.043925335009892784
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,16,balanced,0.06428800026575725
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,16,balanced,0.07180266578992207
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,16,balanced,0.07389866809050243
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,16,balanced,0.07586666444937389
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,16,balanced,0.07619200150171916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,16,balanced,0.11311999956766765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,16,balanced,0.103685329357783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,16,balanced,0.0960746705532074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,16,power_law_1.01,3.5950080871582033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,16,balanced,0.25412799914677936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,16,balanced,0.2264639933904012
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,16,balanced,0.17454934120178223
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,16,balanced,0.12122133374214172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,16,balanced,0.09858133395512898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,16,balanced,0.10723732908566792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,16,balanced,0.11731200416882832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,16,balanced,0.13406399885813394
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,16,balanced,0.15275733669598898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,16,balanced,0.20940800507863364
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,16,balanced,0.2465333342552185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,16,power_law_1.01,5.112831878662109
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,16,balanced,0.35045866171518963
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,16,balanced,0.4549493392308553
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,16,balanced,0.6277653376261393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,16,power_law_1.01,0.291974401473999
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,16,balanced,0.8753653367360433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,16,power_law_1.01,0.3341759920120239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,16,power_law_1.01,0.1360703945159912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,16,power_law_1.01,0.3551487922668457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,16,power_law_1.01,0.14746240377426148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,16,power_law_1.01,0.4073535919189453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,16,power_law_1.01,0.1503167986869812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,16,power_law_1.01,0.48958721160888674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,16,power_law_1.01,0.15192960500717162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,16,power_law_1.01,0.5907264232635498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,16,power_law_1.01,0.8086336135864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,16,power_law_1.01,1.0301888465881348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,16,power_law_1.2,1.7270784378051758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,16,power_law_1.01,1.4451775550842285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,16,power_law_1.01,7.04835205078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,16,power_law_1.01,0.14799360036849976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,16,power_law_1.2,0.1486400008201599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,16,power_law_1.01,0.17111680507659913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,16,power_law_1.2,0.1686527967453003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,16,power_law_1.01,1.7923839569091797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,16,power_law_1.01,0.18183679580688478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,16,power_law_1.2,0.18880000114440917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,16,1,power_law_1.2,1.813222312927246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,16,power_law_1.01,0.1778880000114441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,16,power_law_1.2,0.19114880561828612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,16,power_law_1.01,0.18810880184173584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,16,power_law_1.2,0.12468479871749878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,16,power_law_1.2,0.20019838809967042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,16,power_law_1.01,0.19363199472427367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,16,power_law_1.2,0.1537984013557434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,16,power_law_1.2,0.23158400058746337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,16,power_law_1.01,2.763167953491211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,16,power_law_1.2,0.14481279850006104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,16,power_law_1.2,0.23783679008483888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,16,power_law_1.2,0.15537279844284058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,16,power_law_1.2,0.17574399709701538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,16,power_law_1.2,0.20999040603637695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,16,power_law_1.2,0.21100161075592042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,16,power_law_1.01,3.510598373413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,16,power_law_1.2,0.2128511905670166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,16,power_law_1.2,0.2148672103881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,16,power_law_1.2,0.23077759742736817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,16,power_law_1.01,0.20064001083374022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,16,power_law_1.2,0.2528512001037598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,16,power_law_1.2,0.24058239459991454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,16,power_law_1.2,0.2512128114700317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,16,power_law_1.2,0.2525504112243652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,16,power_law_1.2,0.26483840942382814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,16,power_law_1.2,0.2631808042526245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,16,power_law_1.2,0.2611840009689331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,16,power_law_1.01,5.742918395996094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,16,power_law_1.2,0.3032063961029053
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,16,power_law_1.01,0.05914880037307739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,16,power_law_1.2,0.28574719429016116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,16,power_law_1.01,0.06471040248870849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,16,power_law_1.2,0.358841609954834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,16,power_law_1.01,0.0651199996471405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,16,power_law_1.01,0.2099839925765991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,16,power_law_1.2,0.4055488109588623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,16,power_law_1.01,0.06460800170898437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,16,power_law_1.01,0.21960959434509278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,16,power_law_1.2,0.4214144229888916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,16,power_law_1.01,0.06647040247917176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,16,power_law_1.01,0.23717119693756103
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,16,power_law_1.01,0.06840320229530335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,16,power_law_1.2,0.45866241455078127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,16,power_law_1.01,0.25845119953155515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,16,power_law_1.2,0.314355206489563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,16,power_law_1.01,0.07128959894180298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,16,power_law_1.01,0.289190411567688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,16,power_law_1.2,0.33320960998535154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,16,power_law_1.01,0.07264639735221863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,16,power_law_1.01,0.33182079792022706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,16,power_law_1.01,0.07538560032844543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,16,power_law_1.01,6.8654335021972654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,16,power_law_1.01,0.0953984022140503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,16,power_law_1.01,0.40538878440856935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,16,power_law_1.01,0.09121919870376587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,16,power_law_1.2,0.1369472026824951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,16,power_law_1.01,0.5156352043151855
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,16,power_law_1.01,0.08304640054702758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,16,power_law_1.2,0.13672319650650025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,16,power_law_1.01,0.20659201145172118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,16,power_law_1.01,0.6651072025299072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,16,power_law_1.2,0.15101439952850343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,16,power_law_1.01,0.15934720039367675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,16,power_law_1.2,0.14202239513397216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,16,power_law_1.01,0.13744640350341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,16,power_law_1.2,0.545036792755127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,16,power_law_1.01,0.8585280418395996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,16,power_law_1.2,0.14778239727020265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,16,power_law_1.01,0.10371840000152588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,16,power_law_1.2,0.36971518993377683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,16,power_law_1.2,0.16818560361862184
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,16,power_law_1.01,0.10188159942626954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,16,power_law_1.01,1.1974080085754395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,16,power_law_1.2,0.42429437637329104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,16,power_law_1.2,0.17898240089416503
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,16,power_law_1.01,0.11152000427246093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,16,power_law_1.2,0.18175359964370727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,16,power_law_1.01,0.1245695948600769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,16,power_law_1.2,0.5171840190887451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,16,power_law_1.2,0.18677760362625123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,16,power_law_1.01,1.5477248191833497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,16,power_law_1.2,0.6302527904510498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,16,power_law_1.2,0.19160319566726686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,16,1,power_law_1.2,1.916204833984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,16,power_law_1.2,0.20234239101409912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,16,power_law_1.2,0.843769645690918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,16,power_law_1.2,0.21110401153564454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,16,power_law_1.01,2.2075775146484373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,16,power_law_1.2,0.2222208023071289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,16,power_law_1.2,1.0844736099243164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,16,power_law_1.2,0.24593920707702638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,16,power_law_1.2,0.2703488111495972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,16,power_law_1.2,0.2956543922424316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,16,power_law_1.01,0.15324800014495848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,16,power_law_1.2,0.3474047899246216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,16,power_law_1.01,0.17474559545516968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,16,power_law_1.2,0.4299136161804199
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,16,power_law_1.01,0.24522879123687744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,16,power_law_1.2,0.5211840152740479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,16,power_law_1.2,0.7109055995941163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,16,power_law_1.2,1.547590446472168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,16,power_law_1.2,0.898470401763916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,16,power_law_1.01,3.257344055175781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,16,power_law_1.2,1.2823360443115235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,16,power_law_1.2,1.9832128524780273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,16,power_law_1.2,0.6395584106445312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,16,power_law_1.2,1.6139839172363282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,16,power_law_1.2,0.8917247772216796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,16,power_law_1.01,4.361523056030274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,16,power_law_1.2,0.06048640012741089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,16,power_law_1.2,1.1733247756958007
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,16,power_law_1.2,0.06176000237464905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,16,power_law_1.2,2.566080093383789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,16,power_law_1.2,0.06487039923667907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,16,power_law_1.2,1.5613247871398925
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,16,power_law_1.2,0.06431999802589417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,16,power_law_1.2,3.1136192321777343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,16,power_law_1.01,0.3108992099761963
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,16,power_law_1.2,0.0582144021987915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,16,power_law_1.2,0.06501759886741638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,16,power_law_1.01,0.43640961647033694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,16,power_law_1.2,2.0004928588867186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,16,power_law_1.2,0.07246080040931702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,16,power_law_1.2,0.07414399981498718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,16,power_law_1.01,0.5824639797210693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,16,power_law_1.01,5.970758438110352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,16,power_law_1.2,0.07349119782447815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,16,1,balanced,0.03398933261632919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,16,1,balanced,0.06211733321348826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,16,1,balanced,0.08070933322111766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,16,power_law_1.2,0.0979200005531311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,16,1,balanced,0.10925333698590596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,16,1,balanced,0.17319466670354208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,16,power_law_1.2,0.08880000114440918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,16,power_law_1.2,0.08116480112075805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,16,power_law_1.2,3.873465728759766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,16,power_law_1.2,3.118182373046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,16,power_law_1.2,0.21694719791412354
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,16,power_law_1.2,0.1437183976173401
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,16,power_law_1.2,0.11953279972076417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,16,power_law_1.2,0.10864640474319458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,16,power_law_1.2,0.10236799716949463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,16,power_law_1.2,3.3701503753662108
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,16,power_law_1.01,0.7910079956054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,16,power_law_1.2,0.11550079584121704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,16,1,balanced,0.2103360096613566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,16,1,balanced,0.21132266521453857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,16,power_law_1.2,0.13162879943847655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,16,power_law_1.01,1.1219648361206054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,16,power_law_1.2,0.16225919723510743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,16,1,balanced,0.039488000174363456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,16,1,balanced,0.05038933455944061
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,16,power_law_1.2,0.19467519521713256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,16,power_law_1.2,0.27475199699401853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,16,power_law_1.2,4.500812911987305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,16,power_law_1.2,0.34165120124816895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,16,1,balanced,0.21335999170939127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,16,1,balanced,0.2128373384475708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,16,1,balanced,0.21409600973129272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,16,1,balanced,0.21994666258494058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,16,1,balanced,0.2249600092569987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,16,1,balanced,0.06842133402824402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,16,1,balanced,0.10266666611035664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,16,1,balanced,0.14872533082962036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,16,power_law_1.2,0.480025577545166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,16,1,balanced,0.1944106618563334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,16,1,balanced,0.19640533129374185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,16,1,balanced,0.1970240076382955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,16,1,balanced,0.19917333126068115
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,16,power_law_1.2,0.6645887851715088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,16,1,balanced,0.20061333974202475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,16,1,balanced,0.20356800158818564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,16,1,balanced,0.2092533310254415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,16,1,balanced,0.22236265738805136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,16,power_law_1.2,6.870950317382812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,16,1,balanced,0.2652906576792399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,16,1,balanced,0.27425599098205566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,16,1,balanced,0.27940799792607623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,16,1,balanced,0.28547734022140503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,16,1,balanced,0.2998239994049072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,16,1,balanced,0.3298613429069519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,16,1,balanced,0.35740800698598224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,16,1,balanced,0.5033546686172485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,16,1,balanced,0.22915732860565186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,16,1,balanced,0.23644266525904337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,16,1,balanced,0.24462932348251343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,16,1,balanced,0.25994666417439777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,16,1,balanced,0.2940853238105774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,16,1,balanced,0.32996267080307007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,16,1,balanced,0.46592533588409424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,16,1,balanced,0.5281173388163248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,16,1,balanced,0.7290346622467041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,16,1,balanced,0.9255786736806234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,16,power_law_1.2,3.975507354736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,16,1,balanced,0.559440016746521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,16,1,balanced,1.2566186587015789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,16,1,power_law_1.01,0.032742398977279666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,16,1,balanced,0.7735253175099691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,16,1,power_law_1.01,0.060652798414230345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,16,1,balanced,0.9558506806691488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,16,1,balanced,1.646224021911621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,16,1,power_law_1.01,0.07724159955978394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,16,1,power_law_1.01,0.10085760354995728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,16,1,balanced,1.2807786464691162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,16,1,power_law_1.01,0.11896320581436157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,16,1,balanced,2.372981389363607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,16,1,power_law_1.01,0.17512320280075072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,16,1,power_law_1.01,0.18537600040435792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,16,1,power_law_1.01,0.18714239597320556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,16,1,balanced,3.1006027857462564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,16,1,power_law_1.01,0.1990399956703186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,16,1,power_law_1.01,0.2024384021759033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,16,1,power_law_1.01,0.21597440242767335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,16,1,power_law_1.01,0.24044160842895507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,16,1,power_law_1.01,0.2553024053573608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,16,1,balanced,1.6828907330830891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,16,1,power_law_1.01,0.26480000019073485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,16,1,power_law_1.01,0.27485439777374265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,16,1,power_law_1.01,0.2849024057388306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,16,1,balanced,2.418000062306722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,16,1,power_law_1.01,0.30563840866088865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,16,1,power_law_1.01,0.03736959993839264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,16,1,power_law_1.01,0.3617408037185669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,16,1,power_law_1.01,0.05756800174713135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,16,1,power_law_1.01,0.0704255998134613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,16,1,balanced,3.1587680180867515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,16,1,power_law_1.01,0.09052799940109253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,16,1,power_law_1.2,0.03272959887981415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,16,1,power_law_1.01,0.11020799875259399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,16,1,power_law_1.2,0.06231039762496948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,16,1,power_law_1.01,0.15393279790878295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,16,1,power_law_1.2,0.07670400142669678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,16,1,power_law_1.01,0.1593791961669922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,16,1,power_law_1.2,0.09396479725837707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,16,1,power_law_1.01,0.16100480556488037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,16,1,power_law_1.2,0.11543680429458618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,16,power_law_1.2,9.955897521972656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,16,1,power_law_1.01,0.16543359756469728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,16,1,power_law_1.2,0.17639679908752443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,16,1,power_law_1.01,0.18180480003356933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,16,1,power_law_1.01,0.4370431900024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,16,1,power_law_1.01,0.20370559692382811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,16,1,power_law_1.01,0.21779839992523192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,16,1,power_law_1.01,0.21535999774932862
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,16,power_law_1.2,0.9421567916870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,16,power_law_1.2,6.1983295440673825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,16,1,power_law_1.2,0.03553920090198517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,16,1,power_law_1.2,0.18268799781799316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,16,1,power_law_1.2,0.060083198547363284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,16,1,power_law_1.2,0.19127039909362792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,16,1,power_law_1.01,0.4982143878936768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,16,1,power_law_1.2,0.19921280145645143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,16,power_law_1.2,4.930624008178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,16,1,power_law_1.2,0.20695679187774657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,16,1,power_law_1.01,0.22885119915008545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,16,1,power_law_1.2,0.23269760608673096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,16,1,power_law_1.01,0.23943679332733153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,16,1,power_law_1.2,0.25662078857421877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,16,1,power_law_1.01,0.25039360523223875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,16,1,power_law_1.2,0.25715200901031493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,16,1,power_law_1.01,0.28334081172943115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,16,1,power_law_1.2,0.2674623966217041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,16,1,power_law_1.01,0.33292160034179685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,16,1,power_law_1.2,0.27584640979766845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,16,1,power_law_1.2,0.0707647979259491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,16,1,power_law_1.01,0.38552958965301515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,16,1,power_law_1.2,0.28499839305877683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,16,1,power_law_1.2,0.08886399865150452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,16,1,power_law_1.01,0.5878719806671142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,16,1,power_law_1.01,0.46725120544433596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,16,1,power_law_1.2,0.30687999725341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,16,1,power_law_1.2,0.1068992018699646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,16,1,power_law_1.2,0.36506240367889403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,16,1,power_law_1.2,0.15130879878997802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,16,1,power_law_1.01,0.7725056171417236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,16,1,power_law_1.2,0.15846400260925292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,16,1,power_law_1.2,0.43734397888183596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,16,1,power_law_1.2,0.16126079559326173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,16,1,power_law_1.01,0.9553983688354493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,16,1,power_law_1.2,0.5027200222015381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,16,1,power_law_1.2,0.17685760259628297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,16,1,power_law_1.2,0.19569920301437377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,16,1,power_law_1.01,1.3199359893798828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,16,1,power_law_1.2,0.2055743932723999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,16,1,power_law_1.2,0.5988224029541016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,16,1,power_law_1.2,0.2243648052215576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,16,1,power_law_1.2,0.21780478954315186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,16,1,power_law_1.01,1.68590087890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,16,1,power_law_1.01,0.5603903770446778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,16,1,power_law_1.2,0.2333375930786133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,16,1,power_law_1.2,0.24119040966033936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,16,1,power_law_1.01,0.7375103950500488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,16,1,power_law_1.2,0.258188796043396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,16,1,power_law_1.2,0.7776512145996094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,16,1,power_law_1.2,0.28545920848846434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,16,1,power_law_1.01,2.435251235961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,16,1,power_law_1.2,0.3383039951324463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,16,1,power_law_1.2,0.38874239921569825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,16,1,power_law_1.2,0.46990718841552737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,16,1,power_law_1.2,0.964691162109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,16,1,power_law_1.01,3.186355209350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,16,1,power_law_1.2,0.5695360183715821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,16,balanced,0.07868800063927968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,16,balanced,0.07685333490371704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,16,balanced,0.09107733766237895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,16,balanced,0.10880000392595927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,16,balanced,0.16642133394877115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,16,1,power_law_1.01,0.9269311904907227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,16,1,power_law_1.01,1.294374370574951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,16,1,power_law_1.2,1.3290495872497559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,16,power_law_1.2,7.915277099609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,16,balanced,0.0499946673711141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,16,balanced,0.05402133365472158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,16,balanced,0.08018666505813599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,16,balanced,0.09757866462071736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,16,balanced,0.1341600020726522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,16,1,power_law_1.2,0.7413631916046143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,16,balanced,0.3222666581471761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,16,balanced,0.3927573362986247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,16,balanced,0.5330613454182943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,16,balanced,0.5377333164215088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,16,balanced,0.5437493324279785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,16,balanced,0.547823985417684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,16,balanced,0.5579573313395182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,16,balanced,0.5685919920603434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,16,balanced,0.5782719850540161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,16,balanced,0.5972373485565186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,16,1,power_law_1.01,1.6678207397460938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,16,balanced,0.6438773473103842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,16,balanced,0.6460800170898438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,16,1,power_law_1.2,0.9266624450683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,16,balanced,0.651093324025472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,16,1,power_law_1.01,2.4156543731689455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,16,balanced,0.6586506764094034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,16,balanced,0.6665120124816895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,16,balanced,0.6744693120320638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,16,balanced,0.6150613228480021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,16,balanced,0.6336533228556315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,16,1,power_law_1.01,3.1482175827026366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,16,balanced,0.6685813268025717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,16,balanced,0.7335413297017416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,16,power_law_1.2,1.3168383598327638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,16,balanced,0.8026026884714762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,16,balanced,0.07121066749095917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,16,balanced,0.0749013324578603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,16,balanced,0.09222933650016785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,16,balanced,0.9327253500620524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,16,balanced,0.10313600301742554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,16,balanced,0.1421280006567637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,16,1,power_law_1.2,1.2956928253173827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,16,balanced,1.0714560349782307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,16,balanced,0.2686720093091329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,16,balanced,0.6842400232950846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,16,balanced,0.3884426752726237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,16,balanced,0.7012106577555338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,16,balanced,0.3922400077184041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,16,balanced,1.3374452590942383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,16,balanced,0.396229346593221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,16,balanced,0.40187732378641766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,16,balanced,0.41232534249623615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,16,balanced,1.7556212743123372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,16,balanced,0.41814398765563965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,16,balanced,0.4277600049972534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,16,1,power_law_1.2,1.7049215316772461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,16,balanced,0.44467198848724365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,16,power_law_1.2,6.663302612304688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,16,balanced,2.309605280558268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,16,balanced,0.4620586633682251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,16,balanced,0.47964266935984295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,16,power_law_1.01,0.15247999429702758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,16,balanced,0.5113333463668823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,16,balanced,0.572213331858317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,16,power_law_1.01,0.30670719146728515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,16,balanced,3.0038026173909507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,16,balanced,0.7179199854532877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,16,balanced,0.6341013511021932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,16,balanced,0.7366399765014648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,16,balanced,0.7519199848175049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,16,power_law_1.01,0.3917504072189331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,16,balanced,0.7701173623402914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,16,balanced,0.8716320196787516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,16,balanced,0.8324106534322103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,16,balanced,1.1210933526357014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,16,balanced,4.246143976847331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,16,balanced,0.8960373401641846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,16,balanced,1.0153919855753581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,16,balanced,1.1365013122558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,16,balanced,0.03141866624355316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,16,balanced,0.03335466732581457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,16,balanced,0.034741332133611046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,16,balanced,0.04112533231576284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,16,balanced,1.3785279591878254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,16,balanced,5.680794397989909
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,16,balanced,1.761173407236735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,16,power_law_1.01,0.3845376014709473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,16,power_law_1.01,0.11649279594421387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,16,power_law_1.01,0.3873408079147339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,16,power_law_1.01,0.19827840328216553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,16,balanced,2.2934346199035645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,16,balanced,1.4675359725952148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,16,power_law_1.01,0.2751039981842041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,16,power_law_1.01,0.5493951797485351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,16,balanced,1.9527626037597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,16,balanced,0.06308799982070923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,16,balanced,3.0226720174153647
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,16,balanced,0.09310400485992432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,16,balanced,0.12507200241088867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,16,balanced,0.12550933162371317
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,16,balanced,0.12552533547083536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,16,balanced,0.1262453297773997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,16,balanced,2.550960063934326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,16,1,power_law_1.2,2.452934455871582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,16,balanced,0.12778133153915405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,16,balanced,0.13129066427548727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,16,balanced,0.13409066200256348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,16,balanced,0.1774079998334249
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,16,balanced,0.17096000909805298
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,16,balanced,0.16205333669980368
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,16,balanced,0.3636213143666585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,16,balanced,3.64250119527181
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,16,balanced,0.23957333962122598
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,16,balanced,0.18183465798695883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,16,power_law_1.01,0.5668992042541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,16,balanced,0.2013333241144816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,16,balanced,0.21794132391611734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,16,balanced,0.2569440007209778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,16,power_law_1.01,0.6072447776794434
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,16,balanced,0.3007199962933858
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,16,balanced,0.3859306573867798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,16,balanced,4.860981305440267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,16,power_law_1.01,0.58122239112854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,16,balanced,0.5128000179926554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,16,balanced,4.5082292556762695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,16,balanced,0.6685120264689127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,16,power_law_1.01,0.6304319858551025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,16,power_law_1.01,0.2759552001953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,16,balanced,0.9504960378011068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,16,power_law_1.01,0.6379007816314697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,16,power_law_1.01,0.27419519424438477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,16,power_law_1.01,0.6094272136688232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,16,power_law_1.01,0.41787519454956057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,16,balanced,6.440240224202474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,16,power_law_1.01,0.6050111770629882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,16,power_law_1.01,0.48652801513671873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,16,power_law_1.2,0.15207680463790893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,16,power_law_1.01,0.668991994857788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,16,power_law_1.2,0.29437439441680907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,16,power_law_1.01,0.5067903995513916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,16,power_law_1.01,0.10643199682235718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,16,1,power_law_1.2,1.6683904647827148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,16,power_law_1.01,0.6925759792327881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,16,power_law_1.2,0.34517760276794435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,16,power_law_1.01,0.1818943977355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,16,power_law_1.01,0.522380781173706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,16,power_law_1.01,0.6985663890838623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,16,power_law_1.2,0.3837696075439453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,16,power_law_1.01,0.2525696039199829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,16,power_law_1.01,0.5341631889343261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,16,power_law_1.01,0.7372223854064941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,16,power_law_1.2,0.3920192003250122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,16,power_law_1.01,0.25591681003570554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,16,power_law_1.01,0.5303040027618409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,16,power_law_1.01,0.8284928321838378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,16,power_law_1.2,0.5484672069549561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,16,power_law_1.01,0.5457151889801025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,16,power_law_1.01,0.8802047729492187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,16,power_law_1.2,0.5573503971099854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,16,power_law_1.01,0.5627327919006347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,16,power_law_1.01,0.5580927848815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,16,power_law_1.01,1.052582359313965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,16,power_law_1.2,0.5582015991210938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,16,power_law_1.01,0.6038271903991699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,16,power_law_1.2,0.5393727779388428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,16,power_law_1.01,1.2520256042480469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,16,1,power_law_1.2,3.2044158935546876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,16,power_law_1.01,0.6207935810089111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,16,power_law_1.01,0.275161600112915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,16,power_law_1.2,0.6024640083312989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,16,power_law_1.2,0.11639679670333862
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,16,power_law_1.01,0.05651199817657471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,16,power_law_1.01,0.6552000045776367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,16,power_law_1.01,0.32436480522155764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,16,power_law_1.01,1.6578176498413086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,16,power_law_1.2,0.19687680006027222
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,16,power_law_1.01,0.08481919765472412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,16,power_law_1.01,0.7323391914367676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,16,power_law_1.01,0.34145278930664064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,16,power_law_1.2,0.26620800495147706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,16,power_law_1.01,0.11262719631195069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,16,power_law_1.01,0.8192192077636719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,16,power_law_1.01,0.37467520236968993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,16,power_law_1.2,0.2598655939102173
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,16,power_law_1.01,0.10492160320281982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,16,power_law_1.01,0.3632832050323486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,16,power_law_1.01,0.11389440298080444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,16,power_law_1.2,0.2787775993347168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,16,power_law_1.01,0.1183359980583191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,16,power_law_1.2,0.39493119716644287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,16,power_law_1.01,0.10202879905700683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,16,power_law_1.2,0.6151296138763428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,16,power_law_1.2,0.41901440620422364
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,16,power_law_1.01,0.11598080396652222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,16,power_law_1.2,0.5593472003936768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,16,power_law_1.01,2.00949764251709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,16,power_law_1.2,0.5972544193267822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,16,power_law_1.01,1.0285183906555175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,16,power_law_1.2,0.6694975852966308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,16,power_law_1.01,0.3501055955886841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,16,power_law_1.01,1.2124863624572755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,16,power_law_1.01,2.6737152099609376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,16,power_law_1.2,0.6474559783935547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,16,power_law_1.01,0.3803584098815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,16,power_law_1.2,0.6959487915039062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,16,power_law_1.01,1.5905471801757813
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,16,power_law_1.01,0.10420479774475097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,16,power_law_1.2,0.7492479801177978
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,16,power_law_1.01,0.12120319604873657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,16,power_law_1.01,3.6037696838378905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,16,power_law_1.01,1.9669055938720703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,16,power_law_1.01,0.11292799711227416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,16,power_law_1.01,0.11477119922637939
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,16,power_law_1.01,0.126528000831604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,16,power_law_1.01,0.3915519952774048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,16,power_law_1.01,2.8042367935180663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,16,power_law_1.01,0.12996480464935303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,16,power_law_1.2,0.47267842292785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,16,power_law_1.01,0.4065536022186279
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,16,power_law_1.01,0.13504639863967896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,16,power_law_1.01,5.165824127197266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,16,power_law_1.01,0.4255680084228516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,16,power_law_1.01,0.13747199773788452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,16,power_law_1.2,0.8249728202819824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,16,power_law_1.01,0.27173120975494386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,16,power_law_1.01,0.42952961921691896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,16,power_law_1.01,3.582886505126953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,16,power_law_1.2,0.8860159873962402
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,16,power_law_1.01,0.17328640222549438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,16,power_law_1.01,0.45415678024291994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,16,power_law_1.01,0.1687872052192688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,16,power_law_1.2,1.1156288146972657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,16,power_law_1.01,0.48738560676574705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,16,power_law_1.01,0.19864319562911986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,16,power_law_1.01,0.562553596496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,16,power_law_1.2,1.2715456008911132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,16,power_law_1.2,0.5183551788330079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,16,power_law_1.01,6.597465515136719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,16,power_law_1.01,0.6404799938201904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,16,power_law_1.01,5.048211288452149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,16,power_law_1.2,0.5084735870361328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,16,power_law_1.2,1.7359296798706054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,16,power_law_1.01,0.8003904342651367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,16,power_law_1.2,0.5236224174499512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,16,power_law_1.01,0.9624383926391602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,16,power_law_1.2,2.1517055511474608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,16,power_law_1.2,0.5097343921661377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,16,power_law_1.01,0.2318272113800049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,16,power_law_1.01,1.2751423835754394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,16,power_law_1.2,0.5384640216827392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,16,power_law_1.2,0.5787199974060059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,16,power_law_1.2,2.992064094543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,16,power_law_1.01,1.6224895477294923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,16,power_law_1.01,6.716806030273437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,16,power_law_1.2,0.600761604309082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,16,power_law_1.2,0.6246208190917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,16,power_law_1.01,2.2245952606201174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,16,power_law_1.2,0.6709760189056396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,16,power_law_1.01,0.28296959400177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,16,power_law_1.2,3.7220993041992188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,16,power_law_1.2,0.7532735824584961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,16,power_law_1.01,0.3380160093307495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,16,power_law_1.2,0.1076159954071045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,16,power_law_1.01,2.8790592193603515
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,16,power_law_1.01,0.46491518020629885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,16,power_law_1.2,0.8227071762084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,16,power_law_1.2,0.1817728042602539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,16,power_law_1.01,0.5612544059753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,16,power_law_1.2,0.2412735939025879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,16,power_law_1.2,1.0649727821350097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,16,power_law_1.2,0.25508480072021483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,16,power_law_1.01,0.8279232025146485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,16,power_law_1.2,5.740192031860351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,16,power_law_1.2,1.278656005859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,16,power_law_1.2,0.27151999473571775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,16,power_law_1.01,4.114271926879883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,16,power_law_1.01,1.2429439544677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,16,power_law_1.2,0.3082304000854492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,16,power_law_1.2,1.6777280807495116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,16,1,balanced,0.03566399961709976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,16,1,balanced,0.06649599969387054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,16,1,balanced,0.08164266745249431
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,16,1,balanced,0.10525866349538167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,16,1,balanced,0.1650773286819458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,16,power_law_1.2,0.33279359340667725
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,16,power_law_1.2,0.7810111999511719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,16,1,balanced,0.33027732372283936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,16,1,balanced,0.4505546490351359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,16,1,balanced,0.45070401827494305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,16,power_law_1.2,0.35092480182647706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,16,1,balanced,0.45231465498606366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,16,1,balanced,0.45498132705688477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,16,power_law_1.2,0.35161600112915037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,16,1,balanced,0.45766933759053546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,16,power_law_1.01,5.5523326873779295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,16,1,balanced,0.4612319866816203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,16,1,balanced,0.46454934279123944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,16,power_law_1.2,0.34764800071716306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,16,power_law_1.2,0.05619199872016907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,16,1,balanced,0.4710613489151001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,16,1,balanced,0.4782079855600993
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,16,power_law_1.2,0.08471680283546448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,16,power_law_1.2,0.3624511957168579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,16,1,balanced,0.4851306676864624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,16,1,balanced,0.6403466860453287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,16,power_law_1.2,0.11163519620895386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,16,1,balanced,0.03977066775163015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,16,power_law_1.2,0.39957120418548586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,16,1,balanced,0.6737653414408366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,16,power_law_1.2,2.1337087631225584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,16,power_law_1.2,0.11090559959411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,16,1,balanced,0.7043840090433756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,16,power_law_1.2,0.4033664226531982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,16,1,balanced,0.7657866477966309
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,16,power_law_1.2,0.1131327986717224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,16,power_law_1.2,6.82252197265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,16,power_law_1.2,0.40659198760986326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,16,1,balanced,0.04827733337879181
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,16,power_law_1.2,0.10940799713134766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,16,1,balanced,0.052015999952952065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,16,power_law_1.2,2.9243263244628905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,16,1,balanced,0.06492800017197926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,16,1,balanced,0.07896533111731212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,16,1,balanced,0.12194666266441345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,16,1,balanced,0.20333866278330484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,16,1,balanced,0.3124000032742818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,16,1,balanced,0.06570666531721751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,16,1,balanced,0.3129066626230876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,16,1,balanced,0.3131413261095683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,16,1,balanced,0.3169599970181783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,16,1,balanced,0.32230933507283527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,16,1,balanced,0.32210665941238403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,16,power_law_1.2,3.7945343017578126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,16,1,balanced,0.325381338596344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,16,1,balanced,0.8324213027954102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,16,1,balanced,0.32949866851170856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,16,1,balanced,0.33666133880615234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,16,1,balanced,0.9692959785461426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,16,power_law_1.2,0.4360447883605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,16,1,balanced,1.2793440024058025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,16,power_law_1.2,0.44345598220825194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,16,1,balanced,1.530314604441325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,16,power_law_1.2,0.4843584060668945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,16,1,balanced,1.987866719563802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,16,power_law_1.2,0.5719871997833252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,16,power_law_1.2,5.691667175292968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,16,power_law_1.2,0.6585216045379638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,16,1,balanced,0.34462400277455646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,16,1,balanced,2.7215681076049805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,16,1,balanced,0.3325653274854024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,16,power_law_1.2,0.07934719920158387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,16,1,balanced,0.3526346683502197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,16,power_law_1.2,0.8500543594360351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,16,1,balanced,0.45258665084838867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,16,1,balanced,0.07657066484292348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,16,power_law_1.2,0.1119871973991394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,16,1,balanced,0.09912533561388652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,16,1,balanced,0.14633066455523172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,16,1,balanced,0.2659146587053935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,16,1,balanced,3.8452320098876953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,16,1,balanced,0.3795200188954671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,16,1,balanced,0.381168007850647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,16,1,power_law_1.01,0.03504000008106232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,16,1,balanced,0.3808159828186035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,16,1,balanced,0.3837706645329793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,16,1,power_law_1.01,0.06080639958381653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,16,1,balanced,0.38862931728363037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,16,1,balanced,0.39828264713287354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,16,power_law_1.2,7.346649932861328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,16,1,power_law_1.01,0.07957760095596314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,16,1,balanced,0.40250666936238605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,16,1,power_law_1.01,0.10375039577484131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,16,1,power_law_1.2,2.4131071090698244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,16,1,balanced,0.4028640190760295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,16,1,balanced,0.5011839866638184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,16,1,power_law_1.01,0.15585919618606567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,16,power_law_1.2,0.9928128242492675
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,16,power_law_1.2,0.10004479885101318
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,16,1,balanced,0.03140799949566523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,16,1,power_law_1.01,0.21537280082702637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,16,power_law_1.2,0.1128383994102478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,16,1,power_law_1.01,0.25829761028289794
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,16,power_law_1.2,0.11533440351486206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,16,1,power_law_1.01,0.35916800498962403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,16,power_law_1.2,0.113811194896698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,16,1,power_law_1.01,0.03860479891300202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,16,1,balanced,0.4124480088551839
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,16,power_law_1.2,0.12410240173339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,16,1,balanced,0.42060800393422443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,16,1,balanced,0.7048906485239664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,16,1,balanced,0.032314665615558624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,16,1,balanced,0.036890665690104164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,16,power_law_1.2,0.1460927963256836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,16,1,balanced,0.5295306841532389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,16,power_law_1.2,1.3232640266418456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,16,1,balanced,0.8388693332672119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,16,1,balanced,0.5602933168411255
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,16,power_law_1.2,0.13966720104217528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,16,1,balanced,0.5967040061950684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,16,1,balanced,1.2593386967976887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,16,power_law_1.2,0.1338752031326294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,16,1,balanced,0.667685349782308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,16,power_law_1.2,1.651033592224121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,16,1,balanced,0.7390453020731608
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,16,power_law_1.2,0.2875391960144043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,16,1,balanced,1.7315306663513184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,16,1,balanced,0.8864693641662598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,16,1,power_law_1.01,0.04529919922351837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,16,power_law_1.2,0.1636031985282898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,16,1,balanced,1.1817333698272705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,16,1,balanced,2.43668794631958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,16,1,balanced,0.049882665276527405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,16,1,balanced,0.07628799974918365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,16,1,balanced,0.1181599994500478
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,16,power_law_1.2,0.17466880083084108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,16,1,balanced,1.476469357808431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,16,1,balanced,0.16242667039235434
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,16,1,balanced,0.1657386620839437
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,16,power_law_1.2,0.20662400722503663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,16,1,balanced,1.9131253560384114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,16,1,balanced,3.3564106623331704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,16,power_law_1.2,2.318297576904297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,16,1,balanced,2.6661760012308755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,16,1,power_law_1.01,0.07278079986572265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,16,1,power_law_1.01,0.0976639986038208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,16,1,power_law_1.01,0.3602880001068115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,16,1,power_law_1.01,0.13842560052871705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,16,power_law_1.2,3.0887359619140624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,16,1,power_law_1.01,0.37637760639190676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,16,power_law_1.2,0.23479039669036866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,16,1,power_law_1.01,0.1893056035041809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,16,1,power_law_1.01,0.39080960750579835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,16,power_law_1.2,0.2908096075057983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,16,1,power_law_1.01,0.244486403465271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,16,1,power_law_1.01,0.40342397689819337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,16,power_law_1.2,0.35727360248565676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,16,1,power_law_1.01,0.27698559761047364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,16,1,power_law_1.01,0.4141183853149414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,16,1,power_law_1.01,0.28718080520629885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,16,power_law_1.2,0.5151872158050537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,16,1,balanced,4.5026505788167315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,16,1,power_law_1.01,0.42870402336120605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,16,power_law_1.2,4.567007827758789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,16,1,balanced,0.1669173240661621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,16,1,power_law_1.01,0.29485440254211426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,16,1,balanced,0.16893333196640015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,16,power_law_1.2,0.6205183982849121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,16,1,balanced,0.1728000044822693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,16,1,power_law_1.01,0.44394879341125487
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,16,1,balanced,0.17748800913492838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,16,1,power_law_1.01,0.3060096025466919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,16,1,balanced,0.18475733200709024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,16,1,balanced,0.19338132937749228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,16,1,power_law_1.01,0.4727424144744873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,16,1,balanced,0.20392000675201416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,16,power_law_1.2,0.9152576446533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,16,1,balanced,0.21171200275421143
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,16,1,balanced,0.2582613428433736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,16,1,power_law_1.01,0.6225088119506836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,16,1,balanced,0.3006826639175415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,16,1,balanced,0.2881279985109965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,16,1,balanced,0.34409066041310626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,16,1,power_law_1.01,0.6510784149169921
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,16,1,balanced,0.35922133922576904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,16,1,balanced,0.4406293233235677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,16,1,balanced,3.5284318923950195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,16,power_law_1.2,6.008358383178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,16,1,balanced,0.5484586556752523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,16,1,power_law_1.01,0.046342399716377256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,16,1,power_law_1.2,0.034585601091384886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,16,1,power_law_1.01,0.05041279792785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,16,1,power_law_1.01,0.3136768102645874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,16,1,power_law_1.2,0.06501759886741638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,16,1,power_law_1.01,0.06267520189285278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,16,1,power_law_1.01,0.3292351961135864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,16,1,power_law_1.2,0.07955840229988098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,16,1,power_law_1.01,0.07819520235061646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,16,1,power_law_1.2,0.1034559965133667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,16,power_law_1.2,1.3910783767700194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,16,1,power_law_1.01,0.11223679780960083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,16,1,power_law_1.2,0.14652160406112671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,16,1,power_law_1.2,0.03948799967765808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,16,1,power_law_1.01,0.6893951892852783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,16,1,power_law_1.01,0.14921599626541138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,16,1,power_law_1.2,0.04624640047550201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,16,1,power_law_1.2,0.1981376051902771
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,16,1,balanced,0.6657439867655436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,16,1,power_law_1.01,0.8341312408447266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,16,1,power_law_1.2,0.07525119781494141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,16,1,power_law_1.2,0.24893438816070557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,16,1,power_law_1.2,0.09755520224571228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,16,1,power_law_1.01,0.9052736282348632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,16,1,power_law_1.2,0.33334400653839114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,16,1,power_law_1.2,0.1321023941040039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,16,1,power_law_1.01,0.36291840076446535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,16,1,power_law_1.01,1.0622719764709472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,16,1,power_law_1.2,0.3578687906265259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,16,1,power_law_1.2,0.17964160442352295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,16,1,power_law_1.01,0.3985408067703247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,16,1,power_law_1.2,0.3750783920288086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,16,1,power_law_1.01,0.4284800052642822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,16,1,power_law_1.2,0.20906879901885986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,16,1,power_law_1.01,0.1721791982650757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,16,1,power_law_1.2,0.3792128086090088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,16,1,power_law_1.01,0.5183296203613281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,16,1,balanced,0.9373653729756674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,16,1,power_law_1.2,0.2760447978973389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,16,1,power_law_1.01,0.20349440574645997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,16,1,power_law_1.2,0.38905599117279055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,16,1,power_law_1.2,0.2836031913757324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,16,1,power_law_1.01,0.21377921104431152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,16,1,power_law_1.2,0.41208319664001464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,16,1,power_law_1.2,0.2884927988052368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,16,1,power_law_1.01,0.2329024076461792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,16,1,power_law_1.01,0.028486400842666626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,16,1,power_law_1.2,0.3043776035308838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,16,1,power_law_1.2,0.4215871810913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,16,1,power_law_1.01,0.23836801052093506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,16,1,power_law_1.01,0.030368000268936157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,16,1,power_law_1.01,1.2400832176208496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,16,1,power_law_1.2,0.3164736032485962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,16,1,power_law_1.2,0.45032958984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,16,1,power_law_1.01,0.26112000942230223
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,16,1,power_law_1.01,0.032742398977279666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,16,1,power_law_1.2,0.33728001117706297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,16,1,power_law_1.2,0.4815360069274902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,16,1,power_law_1.01,1.6358335494995118
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,16,1,power_law_1.01,0.04730879962444305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,16,1,power_law_1.01,0.5565375804901123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,16,1,power_law_1.2,0.3651263952255249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,16,1,power_law_1.2,0.6170495986938477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,16,1,power_law_1.01,0.06833279728889466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,16,1,power_law_1.01,0.6000383853912353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,16,1,power_law_1.2,0.40546560287475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,16,1,power_law_1.2,0.6538496017456055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,16,1,power_law_1.01,0.09493759870529175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,16,1,power_law_1.01,0.6974976062774658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,16,1,power_law_1.2,0.43709440231323243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,16,1,power_law_1.2,0.7154623985290527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,16,1,power_law_1.01,0.1194815993309021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,16,1,power_law_1.01,0.8020416259765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,16,1,power_law_1.01,0.13428479433059692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,16,1,power_law_1.01,0.27815680503845214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,16,1,power_law_1.01,0.9889408111572265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,16,1,balanced,0.9769919713338217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,16,1,power_law_1.01,0.1380031943321228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,16,1,power_law_1.01,0.2944000005722046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,16,1,power_law_1.01,0.14261120557785034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,16,1,power_law_1.01,1.18341121673584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,16,1,power_law_1.01,0.31293439865112305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,16,1,power_law_1.01,2.0202304840087892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,16,1,power_law_1.01,0.15159679651260377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,16,1,power_law_1.01,0.3239871978759766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,16,1,power_law_1.01,0.15768959522247314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,16,1,power_law_1.01,1.5698944091796876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,16,1,power_law_1.2,0.519385576248169
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,16,1,power_law_1.01,0.1702015995979309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,16,1,power_law_1.2,0.8392640113830566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,16,1,power_law_1.01,0.409932804107666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,16,1,power_law_1.2,0.5589568138122558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,16,1,power_law_1.01,2.815007972717285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,16,1,power_law_1.01,1.9586240768432617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,16,1,power_law_1.01,0.47681918144226076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,16,1,power_law_1.2,0.6117695808410645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,16,1,balanced,1.3638985951741536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,16,1,power_law_1.01,0.5101888179779053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,16,1,power_law_1.2,0.7137087821960449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,16,1,power_law_1.01,0.6524159908294678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,16,1,power_law_1.01,2.744063949584961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,16,1,power_law_1.01,3.578246307373047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,16,1,power_law_1.01,0.7915520191192627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,16,1,power_law_1.01,0.1835584044456482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,16,1,power_law_1.01,0.991859245300293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,16,1,power_law_1.01,0.1984063982963562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,16,1,power_law_1.01,3.522777557373047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,16,1,power_law_1.01,1.3226431846618651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,16,1,power_law_1.2,0.04753279983997345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,16,1,power_law_1.2,0.8151488304138184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,16,1,power_law_1.2,0.05126399993896484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,16,1,power_law_1.01,1.8584575653076172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,16,1,power_law_1.2,0.06270080208778381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,16,1,power_law_1.2,1.0102656364440918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,16,1,power_law_1.2,0.0779263973236084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,16,1,power_law_1.2,0.9156864166259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,16,1,power_law_1.2,1.1916352272033692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,16,1,power_law_1.2,0.10559359788894654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,16,1,power_law_1.01,2.4115520477294923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,16,1,power_law_1.2,1.070470428466797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,16,balanced,0.031248000760873158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,16,balanced,0.03160000095764796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,16,balanced,0.04177066683769226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,16,balanced,0.05865600208441416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,16,balanced,0.07243200143178304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,16,1,power_law_1.2,1.5796223640441895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,16,balanced,0.16553599635759988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,16,1,power_law_1.2,1.2678912162780762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,16,1,power_law_1.01,3.5061569213867188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,16,1,power_law_1.2,1.9700799942016602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,16,1,power_law_1.2,1.6527744293212892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,16,1,power_law_1.01,0.21193599700927734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,16,1,power_law_1.2,0.14005119800567628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,16,1,power_law_1.01,0.24113280773162843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,16,1,power_law_1.2,2.0464000701904297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,16,1,power_law_1.2,0.02956160008907318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,16,balanced,0.16869866847991943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,16,1,power_law_1.2,2.7517183303833006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,16,balanced,0.16967467466990152
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,16,1,power_law_1.2,0.031200000643730165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,16,1,power_law_1.01,4.606911849975586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,16,1,power_law_1.2,0.035020801424980166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,16,balanced,0.03544000039498011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,16,balanced,0.03942399968703588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,16,balanced,0.037685332198937736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,16,balanced,0.04420800010363261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,16,1,power_law_1.2,2.822323226928711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,16,1,power_law_1.2,3.5256961822509765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,16,balanced,0.050026665131251015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,16,balanced,0.047824000318845115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,16,balanced,0.05484800040721893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,16,balanced,0.05863999823729197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,16,balanced,0.0787360022465388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,16,balanced,0.1323360006014506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,16,balanced,0.13405332962671915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,16,1,power_law_1.2,3.6133502960205077
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,16,1,power_law_1.2,0.04756479859352112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,16,balanced,0.06400000055631001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,16,balanced,0.16432533661524454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,16,balanced,0.16911466916402182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,16,balanced,0.169813334941864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,16,1,power_law_1.2,0.06625919938087463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,16,balanced,0.1721973419189453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,16,balanced,0.17499732971191406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,16,balanced,0.17909334103266397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,16,1,power_law_1.2,0.09291520118713378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,16,1,power_law_1.01,0.3079744100570679
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,16,1,power_law_1.2,0.11516799926757812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,16,balanced,0.13652799526850382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,16,balanced,0.13979732990264893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,16,balanced,0.14196266730626425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,16,balanced,0.1467626690864563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,16,balanced,0.15243200461069742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,16,balanced,0.1563146710395813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,16,balanced,0.16660267114639282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,16,balanced,0.17747199535369873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,16,balanced,0.1889280080795288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,16,balanced,0.21008533239364624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,16,balanced,0.25061333179473877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,16,balanced,0.02741866558790207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,16,balanced,0.291866660118103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,16,balanced,0.18594666322072348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,16,balanced,0.37355732917785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,16,balanced,0.18835200866063437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,16,1,power_law_1.2,0.16075520515441893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,16,balanced,0.20171733697255453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,16,balanced,0.4538559913635254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,16,balanced,0.20983999967575073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,16,balanced,0.1731839974721273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,16,balanced,0.21973333756128946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,16,balanced,0.1755839983622233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,16,1,power_law_1.01,0.2764415979385376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,16,1,power_law_1.2,0.19979519844055177
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,16,1,power_law_1.2,0.132697594165802
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,16,1,power_law_1.01,0.3554879903793335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,16,1,power_law_1.2,0.21497600078582763
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,16,1,power_law_1.2,0.1378175973892212
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,16,1,power_law_1.01,0.3515455961227417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,16,1,power_law_1.2,0.22427520751953126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,16,1,power_law_1.2,0.1417088031768799
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,16,1,power_law_1.01,0.4610176086425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,16,balanced,0.027642667293548584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,16,1,power_law_1.2,0.23845760822296141
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,16,balanced,0.027877333263556164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,16,1,power_law_1.2,0.15134079456329347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,16,1,power_law_1.01,0.5214719772338867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,16,1,power_law_1.2,0.256876802444458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,16,1,power_law_1.2,0.15792640447616577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,16,balanced,0.23912533124287924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,16,balanced,0.2813226580619812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,16,1,power_law_1.2,0.2796031951904297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,16,1,power_law_1.2,0.16955519914627076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,16,balanced,0.3192746639251709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,16,balanced,0.3996586799621582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,16,1,power_law_1.2,0.30185599327087403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,16,1,power_law_1.2,0.18437119722366332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,16,balanced,0.48495999972025555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,16,balanced,0.6802773475646973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,16,1,power_law_1.2,0.19729280471801758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,16,balanced,0.8385866483052572
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,16,1,power_law_1.2,0.21361279487609863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,16,balanced,0.0295413335164388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,16,balanced,1.190170685450236
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,16,1,power_law_1.2,0.24184958934783934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,16,balanced,0.6370773315429688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,16,balanced,0.1793760061264038
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,16,1,power_law_1.2,0.3102976083755493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,16,balanced,0.1837973395983378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,16,1,power_law_1.01,0.7556352138519287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,16,1,power_law_1.2,3.15930233001709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,16,balanced,0.18897600968678793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,16,balanced,0.19981332619984946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,16,1,power_law_1.2,0.27768959999084475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,16,balanced,0.2095680038134257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,16,power_law_1.01,0.06792320013046264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,16,balanced,0.2199946641921997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,16,balanced,0.2388319969177246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,16,1,power_law_1.2,0.3601279973983765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,16,power_law_1.01,0.09978240132331848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,16,balanced,0.2760213414827983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,16,1,power_law_1.2,0.3157439947128296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,16,balanced,0.3156213363011678
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,16,1,power_law_1.2,0.35320320129394533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,16,power_law_1.01,0.10354559421539307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,16,balanced,0.39474133650461835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,16,1,power_law_1.2,0.3285183906555176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,16,balanced,0.4761546850204468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,16,power_law_1.01,0.10486400127410889
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,16,1,power_law_1.2,0.4661759853363037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,16,balanced,0.03348266581694285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,16,balanced,0.6943039894104004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,16,balanced,0.05198400219281515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,16,power_law_1.01,0.10503679513931274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,16,1,power_law_1.2,0.41353597640991213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,16,balanced,0.8465119997660319
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,16,1,power_law_1.2,0.5305024147033691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,16,power_law_1.01,0.15802240371704102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,16,balanced,0.7921013037363688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,16,1,power_law_1.2,0.48538880348205565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,16,balanced,1.1927413145701091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,16,power_law_1.01,0.16419199705123902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,16,balanced,1.1286880175272624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,16,1,power_law_1.2,0.7135615825653077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,16,1,power_law_1.2,0.530463981628418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,16,1,power_law_1.01,0.756876802444458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,16,balanced,1.4674506187438965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,16,1,power_law_1.2,0.6595327854156494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,16,1,power_law_1.2,0.7684991836547852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,16,1,power_law_1.01,1.0381312370300293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,16,balanced,2.1362826029459634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,16,1,power_law_1.2,0.8036735534667969
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,16,balanced,0.05427733560403188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,16,balanced,0.05486399928728739
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,16,1,power_law_1.2,1.0507519721984864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,16,balanced,0.05608533322811127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,16,balanced,0.05608533322811127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,16,balanced,0.05638400216897329
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,16,balanced,0.07654400169849396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,16,balanced,2.8089119593302407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,16,balanced,0.07229866584142049
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,16,balanced,0.06642666459083557
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,16,balanced,0.13926399747530618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,16,power_law_1.01,0.06164479851722717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,16,balanced,0.14448533455530801
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,16,1,power_law_1.2,1.4318400382995606
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,16,balanced,0.10803733269373576
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,16,balanced,0.0751200020313263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,16,balanced,0.08057066798210144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,16,power_law_1.01,0.060652798414230345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,16,power_law_1.01,0.09326720237731934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,16,power_law_1.01,0.0939903974533081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,16,balanced,1.5339093208312988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,16,power_law_1.01,0.09318400025367737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,16,1,power_law_1.2,1.0152447700500489
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,16,1,power_law_1.01,1.4013248443603517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,16,power_law_1.01,0.10022399425506592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,16,balanced,2.2311360041300454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,16,balanced,1.5248746871948242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,16,power_law_1.01,0.09390720129013061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,16,1,power_law_1.2,1.3492480278015138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,16,power_law_1.01,0.12785279750823975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,16,balanced,0.08845866719881694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,16,balanced,2.243552049001058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,16,balanced,2.9247945149739585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,16,power_law_1.01,0.13167359828948974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,16,power_law_1.2,0.06729599833488464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,16,1,power_law_1.2,1.8949951171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,16,power_law_1.01,0.1322175979614258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,16,power_law_1.2,0.09671040177345276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,16,balanced,3.0207306543986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,16,power_law_1.01,0.13377280235290528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,16,power_law_1.2,0.05937280058860779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,16,power_law_1.01,0.165664005279541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,16,power_law_1.01,0.13443200588226317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,16,power_law_1.2,0.09256319999694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,16,power_law_1.01,0.16906239986419677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,16,power_law_1.01,0.13929599523544312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,16,power_law_1.2,0.085971200466156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,16,power_law_1.01,0.17043839693069457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,16,power_law_1.01,0.09247999787330627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,16,power_law_1.2,0.0949504017829895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,16,power_law_1.01,0.17519359588623046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,16,power_law_1.01,0.09653760194778442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,16,power_law_1.2,0.09031680226325989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,16,power_law_1.01,0.17968640327453614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,16,power_law_1.01,0.09388160109519958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,16,power_law_1.2,0.15998719930648803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,16,power_law_1.01,0.18380160331726075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,16,power_law_1.01,0.029100799560546876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,16,power_law_1.2,0.09702399969100953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,16,power_law_1.2,0.16350719928741456
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,16,power_law_1.01,0.04622719883918762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,16,power_law_1.2,0.10312319993972778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,16,power_law_1.2,0.1641983985900879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,16,1,power_law_1.2,2.4480703353881834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,16,power_law_1.2,0.0986624002456665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,16,power_law_1.2,0.16798720359802247
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,16,balanced,0.09802132844924927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,16,power_law_1.2,0.15957119464874267
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,16,balanced,0.11762133240699768
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,16,balanced,0.13901333014170328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,16,power_law_1.2,0.16521600484848023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,16,balanced,0.18920000394185385
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,16,balanced,0.2332693338394165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,16,power_law_1.2,0.16213120222091676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,16,balanced,0.32895465691884357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,16,power_law_1.01,0.15751680135726928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,16,power_law_1.2,0.16852480173110962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,16,power_law_1.01,0.19448959827423096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,16,power_law_1.2,0.1691007971763611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,16,power_law_1.01,0.20517120361328126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,16,power_law_1.2,0.17505919933319092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,16,power_law_1.01,0.04661119878292084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,16,power_law_1.2,0.179750394821167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,16,power_law_1.01,0.04658559858798981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,16,power_law_1.01,0.14752639532089235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,16,power_law_1.2,0.17068159580230713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,16,power_law_1.2,0.18637440204620362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,16,power_law_1.01,0.1505728006362915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,16,power_law_1.2,0.1948032021522522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,16,power_law_1.01,0.159660804271698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,16,power_law_1.2,0.20862081050872802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,16,power_law_1.01,0.17445119619369506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,16,power_law_1.2,0.22140800952911377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,16,1,power_law_1.2,3.5398784637451173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,16,power_law_1.01,0.18234879970550538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,16,power_law_1.2,0.25724799633026124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,16,power_law_1.01,0.165721595287323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,16,power_law_1.01,0.21096959114074706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,16,power_law_1.2,0.317958402633667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,16,power_law_1.01,0.21653759479522705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,16,power_law_1.01,0.2588288068771362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,16,power_law_1.2,0.38483200073242185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,16,power_law_1.01,0.3087935924530029
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,16,power_law_1.01,0.046214398741722104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,16,power_law_1.01,0.4026944160461426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,16,power_law_1.01,0.03861120045185089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,16,power_law_1.01,0.4984255790710449
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,16,balanced,0.42634133497873944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,16,power_law_1.01,0.6871935844421386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,16,power_law_1.2,0.060678398609161376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,16,power_law_1.01,0.16643199920654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,16,power_law_1.2,0.09334400296211243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,16,power_law_1.01,0.8580863952636719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,16,power_law_1.01,0.2403264045715332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,16,power_law_1.2,0.08895360231399536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,16,power_law_1.01,0.3097791910171509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,16,power_law_1.2,0.17512960433959962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,16,power_law_1.2,0.09177600145339966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,16,power_law_1.01,1.2198911666870118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,16,power_law_1.01,0.3709887981414795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,16,power_law_1.2,0.1818176031112671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,16,power_law_1.2,0.09162240028381348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,16,1,power_law_1.2,4.641222381591797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,16,power_law_1.2,0.1865280032157898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,16,power_law_1.01,0.46837759017944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,16,power_law_1.2,0.12444800138473511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,16,power_law_1.2,0.1965824007987976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,16,power_law_1.01,1.5889023780822753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,16,power_law_1.01,0.5503744125366211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,16,power_law_1.2,0.1260800004005432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,16,power_law_1.2,0.13129600286483764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,16,power_law_1.01,0.7557248115539551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,16,power_law_1.01,0.1691200017929077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,16,power_law_1.2,0.13325439691543578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,16,power_law_1.01,0.17049599885940553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,16,power_law_1.01,2.33123836517334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,16,power_law_1.01,0.9487360000610352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,16,power_law_1.2,0.13574399948120117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,16,power_law_1.01,0.1764799952507019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,16,power_law_1.2,0.13842560052871705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,16,power_law_1.01,0.18126720190048218
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,16,power_law_1.01,0.046963199973106384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,16,power_law_1.01,1.32608642578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,16,power_law_1.01,0.1863935947418213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,16,power_law_1.01,0.05258240103721619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,16,power_law_1.01,0.19678720235824584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,16,power_law_1.2,0.2089087963104248
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,16,power_law_1.01,0.04912000000476837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,16,power_law_1.01,3.1754688262939452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,16,power_law_1.01,0.20539519786834717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,16,1,balanced,0.02515733242034912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,16,1,balanced,0.025402667621771496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,16,power_law_1.01,0.05234559774398804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,16,1,balanced,0.031514666974544525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,16,power_law_1.01,0.21743359565734863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,16,power_law_1.01,0.051526397466659546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,16,power_law_1.2,0.029766398668289184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,16,power_law_1.01,0.2421247959136963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,16,power_law_1.2,0.4769792079925537
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,16,power_law_1.01,0.06427519917488098
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,16,power_law_1.2,0.0460671991109848
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,16,power_law_1.01,0.06581760048866273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,16,power_law_1.2,0.5883135795593262
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,16,power_law_1.2,0.043884798884391785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,16,power_law_1.01,0.06176000237464905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,16,power_law_1.2,0.7772992134094239
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,16,power_law_1.01,0.11543680429458618
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,16,power_law_1.01,0.11663999557495117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,16,power_law_1.2,0.9875904083251953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,16,power_law_1.01,1.715225601196289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,16,power_law_1.01,0.08282880187034607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,16,1,balanced,0.05207466582457224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,16,1,balanced,0.07976533472537994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,16,power_law_1.01,0.07307519912719726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,16,power_law_1.2,1.4229632377624513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,16,power_law_1.01,0.07685760259628296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,16,power_law_1.01,0.30862720012664796
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,16,power_law_1.01,0.08955519795417785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,16,power_law_1.01,2.489894485473633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,16,power_law_1.01,0.10183680057525635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,16,power_law_1.2,0.14699519872665406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,16,power_law_1.2,1.8293632507324218
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,16,power_law_1.01,0.13393919467926024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,16,power_law_1.2,0.0470335990190506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,16,power_law_1.2,0.14852479696273804
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,16,power_law_1.01,0.15752320289611815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,16,power_law_1.2,0.2206592082977295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,16,power_law_1.01,0.21407361030578614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,16,power_law_1.2,0.2571647882461548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,16,power_law_1.01,3.3512832641601564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,16,1,balanced,0.1258080005645752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,16,power_law_1.2,2.7027776718139647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,16,1,balanced,0.02940800040960312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,16,1,balanced,0.1258240044116974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,16,1,balanced,0.029653333127498627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,16,1,balanced,0.12957866986592612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,16,power_law_1.2,0.31289598941802976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,16,1,balanced,0.13057600458463034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,16,1,balanced,0.12980799873669943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,16,1,balanced,0.13054933150609335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,16,1,balanced,0.13507733742396036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,16,1,balanced,0.13502933581670126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,16,power_law_1.01,0.3619391918182373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,16,1,balanced,0.139765332142512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,16,1,balanced,0.17522132396697998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,16,1,balanced,0.17931199073791504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,16,power_law_1.01,0.4491136074066162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,16,power_law_1.2,3.5122753143310548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,16,power_law_1.01,0.5531583786010742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,16,1,power_law_1.01,0.021721599996089934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,16,power_law_1.01,0.7348159790039063
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,16,power_law_1.01,0.2763776063919067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,16,1,balanced,0.031248000760873158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,16,1,balanced,0.04789866507053375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,16,1,balanced,0.07452799876530965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,16,power_law_1.01,0.3834431886672974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,16,1,balanced,0.12569066882133484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,16,power_law_1.2,0.3677311897277832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,16,power_law_1.01,0.5481472015380859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,16,power_law_1.2,0.45708160400390624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,16,1,balanced,0.18729066848754883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,16,1,balanced,0.20127467314402261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,16,1,balanced,0.21653334299723306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,16,power_law_1.2,0.04270080029964447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,16,1,balanced,0.24675732851028442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,16,1,balanced,0.2775893410046895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,16,power_law_1.2,0.041945600509643556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,16,1,balanced,0.38202134768168133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,16,1,power_law_1.01,0.023686400055885314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,16,1,balanced,0.43961067994435626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,16,power_law_1.2,0.047891199588775635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,16,1,balanced,0.6032213370005289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,16,power_law_1.01,0.9434880256652832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,16,1,balanced,0.7564106782277426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,16,1,balanced,0.12437333663304646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,16,1,balanced,0.1264959971110026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,16,1,balanced,0.12569066882133484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,16,1,balanced,1.0891893704732258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,16,1,balanced,0.12718933820724487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,16,1,balanced,0.12956800063451132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,16,power_law_1.01,1.3199872016906737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,16,1,balanced,0.1311199963092804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,16,1,balanced,0.13194132844607034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,16,1,balanced,0.13685333728790283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,16,power_law_1.2,0.16254719495773315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,16,1,balanced,0.16058133045832315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,16,1,balanced,0.16462399562199911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,16,1,balanced,0.17409066359202066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,16,1,balanced,0.18949333826700845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,16,power_law_1.01,1.7226495742797852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,16,1,balanced,0.20566932360331217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,16,1,balanced,0.23867199818293253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,16,1,balanced,0.27561599016189575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,16,1,power_law_1.01,0.04044800102710724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,16,1,balanced,0.37504533926645917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,16,power_law_1.2,0.052908802032470705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,16,1,power_law_1.01,0.051948797702789304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,16,power_law_1.2,0.048614400625228885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,16,1,power_law_1.01,0.06798080205917359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,16,power_law_1.01,2.4993215560913087
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,16,power_law_1.2,0.05237119793891907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,16,1,power_law_1.01,0.08293120265007019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,16,1,power_law_1.01,0.026476800441741943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,16,power_law_1.2,0.052262401580810545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,16,1,balanced,1.4236639340718586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,16,power_law_1.2,0.5662335872650146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,16,1,power_law_1.01,0.10232959985733033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,16,1,power_law_1.01,0.028249600529670717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,16,1,power_law_1.2,0.022092799842357635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,16,power_law_1.2,0.17373440265655518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,16,1,power_law_1.01,0.10442880392074586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,16,power_law_1.2,0.8056896209716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,16,power_law_1.2,0.1870527982711792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,16,1,power_law_1.01,0.10810879468917847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,16,power_law_1.01,3.364255905151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,16,power_law_1.2,0.21146879196166993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,16,1,power_law_1.01,0.11143039464950562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,16,power_law_1.2,1.0260671615600585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,16,1,power_law_1.2,0.02487040013074875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,16,1,balanced,0.4434239864349365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,16,power_law_1.2,0.26686720848083495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,16,1,power_law_1.2,0.02698880136013031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,16,power_law_1.2,0.311244797706604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,16,1,power_law_1.2,0.04042240083217621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,16,power_law_1.2,1.4125184059143066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,16,power_law_1.2,0.40483841896057127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,16,1,power_law_1.2,0.04932479858398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,16,power_law_1.2,0.06990079879760742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,16,power_law_1.2,0.5081727981567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,16,power_law_1.2,0.06584960222244263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,16,1,power_law_1.2,0.023731200397014617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,16,power_law_1.2,1.9601663589477538
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,16,power_law_1.2,0.05594879984855652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,16,power_law_1.2,0.6998271942138672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,16,1,power_law_1.2,0.040505599975585935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,16,power_law_1.2,0.10485759973526002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,16,1,power_law_1.2,0.05070719718933105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,16,1,power_law_1.01,0.11621760129928589
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,16,power_law_1.2,0.11518080234527588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,16,power_law_1.2,0.9163711547851563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,16,1,power_law_1.2,0.06469759941101075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,16,1,power_law_1.01,0.12272640466690063
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,16,power_law_1.2,0.08138239979743958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,16,power_law_1.2,2.71081600189209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,16,1,power_law_1.01,0.12679680585861205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,16,1,power_law_1.2,0.08043519854545593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,16,power_law_1.2,0.07538560032844543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,16,power_law_1.2,1.2728063583374023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,16,1,power_law_1.01,0.1383679986000061
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,16,power_law_1.2,0.0803712010383606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,16,1,power_law_1.2,0.1015936017036438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,16,1,power_law_1.01,0.16931840181350707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,16,power_law_1.2,0.09283199906349182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,16,1,power_law_1.2,0.061990398168563846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,16,1,power_law_1.2,0.10517120361328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,16,1,power_law_1.01,0.17224960327148436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,16,1,power_law_1.01,0.04097920060157776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,16,power_law_1.2,0.10623359680175781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,16,1,power_law_1.2,0.10689280033111573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,16,1,power_law_1.2,0.0780672013759613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,16,power_law_1.2,1.6476415634155273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,16,1,power_law_1.01,0.18257919549942017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,16,1,power_law_1.01,0.04912000000476837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,16,power_law_1.2,0.13536640405654907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,16,1,power_law_1.2,0.11080319881439209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,16,1,power_law_1.01,0.19901440143585206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,16,1,power_law_1.2,0.09565439820289612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,16,power_law_1.2,3.597542572021484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,16,power_law_1.2,0.17395199537277223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,16,1,power_law_1.2,0.11911040544509888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,16,balanced,0.031328000128269196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,16,1,power_law_1.01,0.22168960571289062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,16,1,power_law_1.2,0.09960319995880126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,16,1,balanced,0.6135679880777994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,16,1,power_law_1.2,0.12456320524215699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,16,1,power_law_1.01,0.27191040515899656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,16,1,power_law_1.2,0.1302016019821167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,16,power_law_1.2,2.6985855102539062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,16,1,power_law_1.01,0.31486079692840574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,16,1,power_law_1.2,0.14075520038604736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,16,1,power_law_1.2,0.16907520294189454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,16,1,power_law_1.01,0.3887295961380005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,16,1,power_law_1.2,0.17359360456466674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,16,1,power_law_1.01,0.4630335807800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,16,1,power_law_1.2,0.18306560516357423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,16,1,power_law_1.2,0.20259199142456055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,16,1,power_law_1.01,0.6258111953735351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,16,power_law_1.2,3.289651107788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,16,1,power_law_1.2,0.2244096040725708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,16,balanced,0.03576533248027166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,16,balanced,0.03156266609827677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,16,balanced,0.035642666121323906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,16,balanced,0.041797334949175514
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,16,power_law_1.2,0.22869760990142823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,16,1,power_law_1.01,0.7911935806274414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,16,1,power_law_1.2,0.2837503910064697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,16,1,power_law_1.2,0.10181119441986083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,16,1,balanced,0.7774453163146973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,16,power_law_1.2,0.29804799556732176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,16,1,power_law_1.2,0.3173759937286377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,16,1,power_law_1.01,1.115884780883789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,16,1,balanced,1.1182879606882732
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,16,power_law_1.2,0.39797120094299315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,16,1,balanced,1.4583892822265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,16,power_law_1.2,0.6041152000427246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,16,1,power_law_1.01,1.4389439582824708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,16,power_law_1.01,0.06723200082778931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,16,1,power_law_1.01,0.06347519755363465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,16,power_law_1.01,0.09991679787635803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,16,balanced,0.037776000797748566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,16,balanced,0.058917333682378135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,16,balanced,0.04807466765244802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,16,balanced,0.07235733171304067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,16,balanced,0.06604266663392384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,16,balanced,0.16461867094039917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,16,power_law_1.01,0.10113279819488526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,16,balanced,0.16963734229405722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,16,balanced,0.17038400967915854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,16,1,power_law_1.2,0.10412160158157349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,16,balanced,0.17734932899475098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,16,1,power_law_1.2,0.1093824028968811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,16,power_law_1.01,0.1050495982170105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,16,balanced,0.1777013341585795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,16,balanced,0.1788426637649536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,16,1,power_law_1.2,0.11749759912490845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,16,balanced,0.18450667460759482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,16,balanced,0.18943999210993448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,16,1,power_law_1.2,0.12933119535446166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,16,balanced,0.20055999358495077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,16,balanced,0.20971733331680298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,16,1,power_law_1.2,0.14010239839553834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,16,balanced,0.21961599588394165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,16,balanced,0.2383520007133484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,16,1,power_law_1.2,0.15637120008468627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,16,balanced,0.2765653332074483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,16,balanced,0.3162026604016622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,16,1,power_law_1.01,0.07983360290527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,16,balanced,0.39558935165405273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,16,balanced,0.164682666460673
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,16,balanced,0.02749866743882497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,16,balanced,0.4758933385213216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,16,balanced,0.027642667293548584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,16,balanced,0.028143999477227528
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,16,balanced,0.029482667644818623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,16,balanced,0.6967786947886149
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,16,balanced,0.03454933315515518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,16,1,power_law_1.2,0.3911295890808105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,16,balanced,0.05341866612434387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,16,balanced,0.05470400055249532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,16,balanced,0.845695972442627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,16,balanced,0.05598400036493937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,16,balanced,0.056602666775385536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,16,balanced,0.05784533421198527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,16,balanced,0.05644266804059347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,16,balanced,1.1929173469543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,16,balanced,0.044154668847719826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,16,balanced,0.07867733140786488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,16,balanced,0.04874666531880697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,16,balanced,0.07438399891058604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,16,balanced,0.053455998500188194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,16,balanced,0.0665226678053538
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,16,balanced,0.14634133378664652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,16,balanced,1.5235999425252278
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,16,balanced,0.14853333433469137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,16,balanced,0.10113599896430969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,16,1,power_law_1.2,0.16045440435409547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,16,1,power_law_1.2,0.17491840124130248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,16,balanced,2.24454402923584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,16,1,power_law_1.2,0.19443199634552003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,16,1,power_law_1.2,0.21925759315490723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,16,1,power_law_1.2,0.2707200050354004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,16,balanced,3.0218025843302407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,16,power_law_1.01,0.10374399423599243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,16,1,power_law_1.2,0.306329607963562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,16,power_law_1.01,0.059699201583862306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,16,balanced,0.05782933533191681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,16,power_law_1.01,0.15595519542694092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,16,1,power_law_1.2,0.39380478858947754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,16,power_law_1.01,0.0932416021823883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,16,power_law_1.01,0.16277120113372803
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,16,balanced,0.07758933305740356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,16,balanced,0.07985066870848338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,16,1,power_law_1.2,0.4757887840270996
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,16,balanced,0.08910399675369263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,16,power_law_1.01,0.16499840021133422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,16,balanced,0.0990613301595052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,16,balanced,0.11858133474985759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,16,balanced,0.1729653278986613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,16,balanced,0.13995200395584106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,16,balanced,0.1704853375752767
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,16,balanced,0.18965333700180054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,16,balanced,0.17269333203633627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,16,balanced,0.2327573299407959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,16,balanced,0.17476266622543335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,16,balanced,0.18032532930374146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,16,balanced,0.32845866680145264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,16,1,power_law_1.2,0.4664639949798584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,16,balanced,0.18518400192260742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,16,balanced,0.18914133310317993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,16,balanced,0.19933332999547324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,16,balanced,0.21056532859802246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,16,1,power_law_1.2,0.6302464008331299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,16,balanced,0.0783733328183492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,16,balanced,0.1297546625137329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,16,balanced,0.13361066579818726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,16,balanced,0.1359946628411611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,16,1,power_law_1.2,0.7904895782470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,16,balanced,0.1400373379389445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,16,power_law_1.01,0.089055997133255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,16,balanced,0.14089600245157877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,16,balanced,0.14593600233395895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,16,balanced,0.15180800358454385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,16,power_law_1.01,0.09455999732017517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,16,balanced,0.1570186714331309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,16,power_law_1.01,0.168230402469635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,16,balanced,0.16673600673675537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,16,power_law_1.01,0.08780159950256347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,16,power_law_1.01,0.17051520347595214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,16,power_law_1.01,0.15843199491500853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,16,power_law_1.01,0.17603839635849
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,16,balanced,0.4259466727574666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,16,power_law_1.01,0.18012800216674804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,16,balanced,0.21811199188232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,16,power_law_1.01,0.18585599660873414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,16,balanced,0.2400373419125875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,16,balanced,0.27974933385849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,16,power_law_1.01,0.196563196182251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,16,balanced,0.31804800033569336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,16,balanced,0.39877867698669434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,16,1,power_law_1.01,0.09660159945487976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,16,balanced,0.4840266704559326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,16,1,power_law_1.2,0.6492288112640381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,16,1,power_law_1.01,0.10019840002059936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,16,balanced,0.6801813443501791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,16,balanced,0.17801600694656372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,16,1,power_law_1.01,0.10292479991912842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,16,1,power_law_1.2,0.8184703826904297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,16,1,power_law_1.01,0.10402560234069824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,16,power_law_1.01,0.16380800008773805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,16,1,power_law_1.01,0.1068160057067871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,16,1,power_law_1.2,1.166047954559326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,16,1,power_law_1.01,0.11356799602508545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,16,1,power_law_1.01,0.1272320032119751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,16,power_law_1.01,0.20633599758148194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,16,1,power_law_1.2,1.5122559547424317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,16,power_law_1.01,0.21551361083984374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,16,1,power_law_1.2,1.116147232055664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,16,power_law_1.01,0.061337602138519284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,16,power_law_1.01,0.2421504020690918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,16,balanced,0.18708799282709757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,16,balanced,0.8368213176727295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,16,balanced,0.20964266856511435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,16,power_law_1.01,0.3003648042678833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,16,balanced,0.2513759930928548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,16,balanced,0.2916959921518962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,16,power_law_1.01,0.36148478984832766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,16,balanced,0.37227733929951984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,16,balanced,0.45401068528493244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,16,power_law_1.01,0.46918401718139646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,16,balanced,0.6347946723302206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,16,power_law_1.01,0.5532671928405761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,16,balanced,0.7906080087025961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,16,1,power_law_1.01,0.13912960290908813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,16,1,power_law_1.01,0.15512959957122802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,16,power_law_1.01,0.7700352191925048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,16,1,power_law_1.01,0.1587391972541809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,16,power_law_1.01,0.09352319836616516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,16,1,power_law_1.01,0.17011840343475343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,16,1,power_law_1.2,1.4491392135620118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,16,1,power_law_1.01,0.19260799884796143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,16,power_law_1.2,0.06771199703216553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,16,1,power_law_1.01,0.2156287908554077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,16,power_law_1.01,0.16500480175018312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,16,power_law_1.2,0.09751039743423462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,16,1,power_law_1.01,0.26012160778045657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,16,power_law_1.01,0.16870399713516235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,16,power_law_1.2,0.09136639833450318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,16,1,power_law_1.01,0.30519039630889894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,16,power_law_1.01,0.17023359537124633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,16,power_law_1.2,0.09965440034866332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,16,balanced,1.128992001215617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,16,power_law_1.01,0.17621760368347167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,16,power_law_1.2,0.09784960150718688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,16,power_law_1.01,0.18048640489578247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,16,power_law_1.01,0.9605952262878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,16,power_law_1.2,0.16088320016860963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,16,power_law_1.01,0.18572800159454345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,16,balanced,1.1867146492004395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,16,power_law_1.01,0.1970047950744629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,16,power_law_1.01,0.20796799659729004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,16,power_law_1.01,0.21667840480804443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,16,power_law_1.01,0.24768640995025634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,16,power_law_1.01,0.030188798904418945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,16,1,power_law_1.01,0.387935996055603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,16,power_law_1.01,0.30395519733428955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,16,power_law_1.01,0.04710400104522705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,16,1,power_law_1.01,0.47199358940124514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,16,power_law_1.01,0.3600703954696655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,16,balanced,1.462272008260091
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,16,power_law_1.01,0.04753279983997345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,16,power_law_1.01,0.09430400133132935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,16,power_law_1.01,0.45427842140197755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,16,1,power_law_1.01,0.6415616035461426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,16,power_law_1.2,0.16224000453948975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,16,power_law_1.01,0.0461760014295578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,16,power_law_1.01,0.0945088028907776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,16,balanced,2.1264373461405435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,16,balanced,1.5315146446228027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,16,power_law_1.2,0.16545920372009276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,16,power_law_1.01,0.047654399275779726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,16,power_law_1.01,0.10073599815368653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,16,power_law_1.2,0.16625280380249025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,16,power_law_1.01,0.043219199776649474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,16,power_law_1.01,0.11948800086975098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,16,power_law_1.2,0.16949119567871093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,16,balanced,2.225210666656494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,16,power_law_1.01,0.04668799936771393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,16,balanced,2.7990185419718423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,16,power_law_1.01,0.12434560060501099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,16,power_law_1.2,0.17417600154876708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,16,power_law_1.01,0.04779520034790039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,16,power_law_1.2,0.06048640012741089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,16,power_law_1.01,0.13062399625778198
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,16,power_law_1.01,0.050860798358917235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,16,power_law_1.2,0.09130240082740784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,16,balanced,2.9185813268025718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,16,power_law_1.01,0.05059199929237366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,16,power_law_1.2,0.09196799993515015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,16,power_law_1.2,0.06101120114326477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,16,power_law_1.01,1.323084831237793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,16,power_law_1.2,0.08960639834403991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,16,1,power_law_1.01,0.8161472320556641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,16,power_law_1.2,0.09255040287971497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,16,power_law_1.2,0.09377279877662659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,16,power_law_1.2,0.09317119717597962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,16,power_law_1.01,1.7141952514648438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,16,1,power_law_1.01,1.157369613647461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,16,power_law_1.2,0.17900799512863158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,16,power_law_1.2,0.18583040237426757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,16,power_law_1.01,0.13415039777755738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,16,1,power_law_1.01,1.5009856224060059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,16,power_law_1.2,0.19623039960861205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,16,power_law_1.01,0.13526400327682495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,16,power_law_1.2,0.20624639987945556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,16,power_law_1.01,0.5509759902954101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,16,power_law_1.01,0.14325120449066162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,16,power_law_1.2,0.2241215944290161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,16,power_law_1.01,0.1479871988296509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,16,power_law_1.01,0.7344319820404053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,16,power_law_1.2,0.24947841167449952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,16,power_law_1.01,0.1510975956916809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,16,power_law_1.2,0.15661439895629883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,16,power_law_1.2,0.32166399955749514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,16,power_law_1.01,0.1595136046409607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,16,power_law_1.01,0.9804672241210938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,16,power_law_1.2,0.163372802734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,16,power_law_1.01,0.17345919609069824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,16,power_law_1.2,0.1655295968055725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,16,power_law_1.01,0.18435839414596558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,16,power_law_1.01,1.3674752235412597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,16,power_law_1.01,0.20949759483337402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,16,power_law_1.2,0.1675968050956726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,16,power_law_1.01,2.546451187133789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,16,power_law_1.01,0.051148802042007446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,16,power_law_1.2,0.16922240257263182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,16,power_law_1.01,0.06579840183258057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,16,power_law_1.2,0.1751039981842041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,16,power_law_1.01,1.7296768188476563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,16,power_law_1.01,0.06276479959487916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,16,power_law_1.2,0.18091520071029663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,16,power_law_1.01,0.05999360084533691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,16,power_law_1.2,0.18520959615707397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,16,power_law_1.2,0.02919679880142212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,16,power_law_1.01,3.311942291259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,16,power_law_1.2,0.3898303985595703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,16,1,balanced,0.02329600105683009
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,16,power_law_1.2,0.04398080110549927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,16,power_law_1.01,2.5668928146362306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,16,power_law_1.2,0.48516478538513186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,16,power_law_1.2,0.04373759925365448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,16,power_law_1.2,0.04652799963951111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,16,power_law_1.01,0.26292479038238525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,16,power_law_1.01,0.3083264112472534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,16,power_law_1.01,0.40343680381774905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,16,power_law_1.01,3.2909568786621093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,16,power_law_1.2,0.09355520009994507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,16,1,balanced,0.027456000447273254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,16,1,balanced,0.029167999823888142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,16,power_law_1.01,0.4946112155914307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,16,1,balanced,0.03532266616821289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,16,power_law_1.2,0.1962623953819275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,16,power_law_1.2,0.09662079811096191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,16,1,balanced,0.025098666548728943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,16,1,balanced,0.03928533444801966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,16,power_law_1.2,0.211244797706604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,16,power_law_1.2,0.11342719793319703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,16,power_law_1.2,0.2216320037841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,16,power_law_1.2,0.1285375952720642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,16,power_law_1.2,0.578873586654663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,16,power_law_1.2,0.2541759967803955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,16,power_law_1.2,0.12663040161132813
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,16,power_law_1.2,0.04301440119743347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,16,power_law_1.2,0.3065727949142456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,16,power_law_1.2,0.7855743885040283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,16,power_law_1.2,0.13038079738616942
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,16,power_law_1.2,0.04289279878139496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,16,power_law_1.2,0.3680959939956665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,16,power_law_1.2,0.13309439420700073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,16,power_law_1.2,1.0044992446899415
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,16,power_law_1.2,0.04655359983444214
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,16,power_law_1.01,0.10918400287628174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,16,power_law_1.2,0.46073598861694337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,16,1,balanced,0.0476746658484141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,16,1,balanced,0.07658666869004567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,16,1,balanced,0.12378133336702983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,16,power_law_1.2,0.0466623991727829
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,16,power_law_1.01,0.1085312008857727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,16,1,balanced,0.12502933541933695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,16,1,balanced,0.12570133805274963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,16,1,balanced,0.12570133805274963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,16,power_law_1.2,1.42042236328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,16,power_law_1.2,0.05018240213394165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,16,1,balanced,0.1274079978466034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,16,1,balanced,0.12764267126719156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,16,1,balanced,0.12982933719952902
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,16,power_law_1.2,0.051507198810577394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,16,1,balanced,0.13186132907867432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,16,1,balanced,0.1360319952170054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,16,1,balanced,0.16086933016777039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,16,power_law_1.2,0.050310397148132326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,16,1,balanced,0.1646880010763804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,16,power_law_1.2,1.905580711364746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,16,1,balanced,0.1729546586672465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,16,1,balanced,0.1911840041478475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,16,1,balanced,0.2076693375905355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,16,1,balanced,0.24035733938217163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,16,1,balanced,0.2758293350537618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,16,power_law_1.2,0.1418496012687683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,16,1,balanced,0.3775893449783325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,16,1,balanced,0.44312000274658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,16,power_law_1.2,0.5618559837341308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,16,power_law_1.2,2.7492799758911133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,16,1,balanced,0.614186684290568
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,16,power_law_1.01,0.07487360239028931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,16,1,balanced,0.7806666692097982
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,16,power_law_1.01,0.07414399981498718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,16,power_law_1.2,0.7789184093475342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,16,1,balanced,1.1215146382649739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,16,power_law_1.2,0.9743488311767579
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,16,power_law_1.2,0.0635968029499054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,16,1,balanced,1.4618719418843586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,16,power_law_1.2,3.553081512451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,16,1,power_law_1.01,0.0219200000166893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,16,power_law_1.2,1.3972543716430663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,16,1,power_law_1.01,0.025286400318145753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,16,1,balanced,0.05027199784914652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,16,1,balanced,0.07991466422875722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,16,1,power_law_1.01,0.027628800272941588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,16,1,balanced,0.12745066483815512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,16,1,balanced,0.12641066312789917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,16,1,balanced,0.12800000111262003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,16,1,power_law_1.01,0.040889599919319154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,16,1,balanced,0.12945600350697836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,16,1,balanced,0.1281599998474121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,16,power_law_1.2,1.8640960693359374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,16,1,balanced,0.12973866860071817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,16,1,power_law_1.01,0.04982399940490723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,16,1,balanced,0.13357333342234293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,16,1,balanced,0.13417067130406699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,16,1,balanced,0.13896532853444418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,16,1,power_law_1.01,0.06436480283737182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,16,1,balanced,0.17525867621103922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,16,1,balanced,0.17858133713404337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,16,1,balanced,0.18780267238616943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,16,1,power_law_1.01,0.08065919876098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,16,1,balanced,0.20160533984502158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,16,1,balanced,0.21766400337219238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,16,1,power_law_1.01,0.09466879963874816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,16,power_law_1.2,2.5801727294921877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,16,power_law_1.01,0.6837632179260253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,16,1,power_law_1.01,0.10076800584793091
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,16,power_law_1.2,0.06382079720497132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,16,1,power_law_1.01,0.1007040023803711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,16,power_law_1.2,0.059520000219345094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,16,power_law_1.01,0.8766271591186523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,16,1,power_law_1.01,0.10395519733428955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,16,power_law_1.2,0.10592639446258545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,16,power_law_1.01,0.07869439721107482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,16,power_law_1.2,0.10236159563064576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,16,power_law_1.01,1.2593215942382812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,16,power_law_1.2,3.5714752197265627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,16,power_law_1.2,0.08344320058822632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,16,1,power_law_1.2,0.021990400552749634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,16,power_law_1.2,0.07534719705581665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,16,1,balanced,0.24623467524846396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,16,power_law_1.01,1.635532760620117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,16,power_law_1.2,0.07930240035057068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,16,1,power_law_1.01,0.02705279886722565
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,16,power_law_1.2,0.0919871985912323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,16,1,power_law_1.01,0.04035199880599975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,16,power_law_1.2,0.10637439489364624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,16,1,power_law_1.01,0.05107839703559876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,16,power_law_1.2,0.1348479986190796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,16,1,power_law_1.01,0.06776959896087646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,16,power_law_1.01,0.09038720130920411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,16,1,power_law_1.01,0.08351359963417053
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,16,power_law_1.01,0.1025920033454895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,16,1,power_law_1.01,0.10130560398101807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,16,1,power_law_1.2,0.02374400049448013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,16,1,power_law_1.01,0.10766719579696656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,16,1,balanced,0.27615465720494586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,16,power_law_1.2,0.14863359928131104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,16,1,power_law_1.2,0.040575999021530154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,16,1,balanced,0.38235731919606525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,16,1,power_law_1.01,0.10817279815673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,16,power_law_1.2,0.15006719827651976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,16,1,power_law_1.2,0.05064319968223572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,16,1,power_law_1.01,0.11012480258941651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,16,power_law_1.2,0.16261119842529298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,16,1,power_law_1.2,0.06565759778022766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,16,1,power_law_1.01,0.11600639820098876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,16,power_law_1.2,0.17581440210342408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,16,1,power_law_1.01,0.1221119999885559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,16,1,power_law_1.2,0.07925119996070862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,16,power_law_1.2,0.18561279773712158
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,16,power_law_1.2,0.16832000017166138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,16,1,power_law_1.01,0.12906240224838256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,16,1,power_law_1.2,0.10051840543746948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,16,power_law_1.2,0.21418240070343017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,16,1,power_law_1.01,0.1387519955635071
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,16,power_law_1.2,0.23150079250335692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,16,1,power_law_1.2,0.10392960309982299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,16,1,power_law_1.01,0.1674623966217041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,16,power_law_1.01,0.12687360048294066
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,16,power_law_1.2,0.31118719577789306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,16,1,power_law_1.2,0.10929280519485474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,16,1,power_law_1.01,0.17178239822387695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,16,power_law_1.01,0.15667200088500977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,16,1,balanced,0.44069333871205646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,16,1,power_law_1.2,0.1106943964958191
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,16,power_law_1.2,0.43331198692321776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,16,power_law_1.01,0.2109312057495117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,16,1,balanced,0.6023626724878947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,16,1,power_law_1.2,0.11568000316619872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,16,1,balanced,0.7580479780832926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,16,power_law_1.01,2.3134719848632814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,16,1,power_law_1.2,0.12114559412002564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,16,power_law_1.2,0.6102911949157714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,16,1,balanced,1.0936319828033447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,16,1,power_law_1.2,0.026579201221466064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,16,1,power_law_1.2,0.028198400139808656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,16,1,balanced,1.4250613848368328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,16,power_law_1.2,0.265228796005249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,16,1,power_law_1.2,0.040633600950241086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,16,balanced,0.035599999129772186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,16,balanced,0.03532800078392029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,16,power_law_1.2,0.32108800411224364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,16,balanced,0.04115733255942663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,16,power_law_1.01,3.0909631729125975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,16,balanced,0.057989334066708885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,16,1,power_law_1.2,0.049747198820114136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,16,balanced,0.03586133321126302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,16,balanced,0.03757333258787791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,16,balanced,0.0376800000667572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,16,balanced,0.04385599990685781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,16,1,power_law_1.2,0.061900800466537474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,16,balanced,0.08059733112653096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,16,balanced,0.09769066174825032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,16,balanced,0.10114133358001709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,16,balanced,0.10302933057149251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,16,1,power_law_1.01,0.10698879957199096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,16,power_law_1.01,0.2661119937896729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,16,balanced,0.10654933253924052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,16,balanced,0.10526933272679646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,16,balanced,0.1092800001303355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,16,1,power_law_1.01,0.112716805934906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,16,balanced,0.11543466647466023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,16,power_law_1.01,0.3966847896575928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,16,balanced,0.11743467052777608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,16,1,power_law_1.01,0.11832319498062134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,16,1,power_law_1.01,0.13916159868240358
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,16,power_law_1.01,0.5424255847930908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,16,1,power_law_1.01,0.15513600111007692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,16,power_law_1.01,0.06199679970741272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,16,1,power_law_1.01,0.15931520462036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,16,balanced,0.09371200203895569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,16,power_law_1.01,0.0723136007785797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,16,balanced,0.10069866975148518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,16,power_law_1.2,0.40954880714416503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,16,balanced,0.09939733147621155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,16,balanced,0.10314133763313293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,16,balanced,0.10321066776911418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,16,balanced,0.10341866811116536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,16,power_law_1.2,0.5076159954071044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,16,1,power_law_1.2,0.13040000200271606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,16,power_law_1.2,0.7112639904022217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,16,1,power_law_1.2,0.14288640022277832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,16,1,power_law_1.2,0.16822400093078613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,16,power_law_1.2,0.881868839263916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,16,1,power_law_1.2,0.1739583969116211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,16,1,power_law_1.2,0.18473600149154662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,16,1,power_law_1.01,0.18283519744873047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,16,power_law_1.2,1.2970751762390136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,16,1,power_law_1.01,0.1712767958641052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,16,power_law_1.01,0.0660863995552063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,16,1,power_law_1.01,0.19939199686050416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,16,1,power_law_1.01,0.19226239919662474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,16,power_law_1.01,0.06922879815101624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,16,1,power_law_1.01,0.21895039081573486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,16,1,power_law_1.01,0.21620481014251708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,16,power_law_1.01,0.09151999950408936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,16,power_law_1.2,1.708902359008789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,16,1,power_law_1.01,0.26954240798950196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,16,balanced,0.12707199652989706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,16,balanced,0.1329919993877411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,16,1,power_law_1.01,0.30490880012512206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,16,balanced,0.14171199997266135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,16,balanced,0.15806933244069418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,16,balanced,0.18701332807540894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,16,1,power_law_1.01,0.3908159971237183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,16,balanced,0.21599467595418295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,16,balanced,0.29517332712809247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,16,1,power_law_1.01,0.4716800212860107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,16,1,power_law_1.2,0.20280320644378663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,16,1,power_law_1.01,0.6432384014129638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,16,1,power_law_1.2,0.07783039808273315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,16,1,power_law_1.01,0.2641151905059814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,16,1,power_law_1.2,0.09488000273704529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,16,power_law_1.01,0.09671040177345276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,16,1,power_law_1.01,0.3158015966415405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,16,1,power_law_1.2,0.09890559911727906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,16,power_law_1.01,0.0975488007068634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,16,1,power_law_1.01,0.3881727933883667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,16,1,power_law_1.2,0.10282880067825317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,16,1,power_law_1.2,0.10469759702682495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,16,balanced,0.3604106505711873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,16,1,power_law_1.2,0.10837759971618652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,16,balanced,0.511738657951355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,16,1,power_law_1.2,0.11378560066223145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,16,balanced,0.1090719997882843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,16,balanced,0.11315733194351196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,16,1,power_law_1.2,0.12739839553833007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,16,balanced,0.11404266953468323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,16,1,power_law_1.2,0.22272639274597167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,16,1,power_law_1.01,0.8119744300842285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,16,1,power_law_1.2,0.14188159704208375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,16,1,power_law_1.2,0.28421759605407715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,16,1,power_law_1.2,0.15671039819717408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,16,power_law_1.01,0.10032639503479004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,16,1,power_law_1.01,1.157107162475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,16,power_law_1.01,0.10150400400161744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,16,power_law_1.2,2.408064079284668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,16,power_law_1.01,0.10194560289382934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,16,power_law_1.01,0.1053056001663208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,16,1,power_law_1.01,1.5035264015197753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,16,power_law_1.01,0.10915199518203736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,16,balanced,0.6385600169499716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,16,power_law_1.01,0.05353599786758423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,16,power_law_1.01,0.11236480474472046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,16,balanced,0.9265173276265463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,16,balanced,0.12341333429018657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,16,power_law_1.01,0.12319999933242798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,16,balanced,0.13290133078893027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,16,balanced,0.13849066694577536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,16,balanced,1.1986613273620605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,16,power_law_1.2,3.3011711120605467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,16,power_law_1.2,0.06206079721450806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,16,1,power_law_1.2,0.3174783945083618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,16,balanced,1.7575732866923015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,16,power_law_1.2,0.06543999910354614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,16,1,power_law_1.01,0.465715217590332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,16,1,power_law_1.2,0.3894207954406738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,16,power_law_1.2,0.07163519859313965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,16,power_law_1.2,0.07241600155830383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,16,1,power_law_1.01,0.6251264095306397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,16,balanced,2.313999970753988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,16,power_law_1.2,0.08972799777984619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,16,power_law_1.2,0.05324800014495849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,16,power_law_1.2,0.0955136001110077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,16,1,power_law_1.01,0.790009593963623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,16,power_law_1.01,0.06320639848709106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,16,power_law_1.2,0.06204800009727478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,16,power_law_1.2,0.09889280200004577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,16,power_law_1.01,0.06318719983100891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,16,power_law_1.01,0.13056639432907105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,16,balanced,0.1564640005429586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,16,balanced,0.18755733966827393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,16,power_law_1.01,0.06379520297050476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,16,balanced,0.21622933944066366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,16,1,power_law_1.2,0.16101759672164917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,16,balanced,0.29309332370758057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,16,power_law_1.01,0.0800000011920929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,16,balanced,0.35545066992441815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,16,1,power_law_1.2,0.17315839529037474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,16,power_law_1.01,0.09667199850082397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,16,balanced,0.5180480082829794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,16,1,power_law_1.2,0.19480960369110106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,16,balanced,0.6467946767807007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,16,1,power_law_1.2,0.46586880683898924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,16,1,power_law_1.2,0.21943039894104005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,16,balanced,0.9385920365651449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,16,1,power_law_1.2,0.629369592666626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,16,1,power_law_1.2,0.26575360298156736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,16,balanced,1.204591989517212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,16,power_law_1.2,0.06481919884681701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,16,power_law_1.2,0.0998207986354828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,16,1,power_law_1.2,0.7916224002838135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,16,1,power_law_1.01,1.1122112274169922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,16,power_law_1.2,0.10243200063705445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,16,power_law_1.2,0.10160000324249267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,16,1,power_law_1.2,1.1162943840026855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,16,power_law_1.2,0.1062656044960022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,16,power_law_1.01,0.09826560020446777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,16,power_law_1.01,0.10149120092391968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,16,1,power_law_1.2,1.447923183441162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,16,1,balanced,0.023333333432674408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,16,1,balanced,0.027471999327341717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,16,1,balanced,0.03575466573238373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,16,1,balanced,0.0664160003264745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,16,1,balanced,0.10548266768455505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,16,1,power_law_1.2,0.30636799335479736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,16,power_law_1.01,0.14074239730834961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,16,1,power_law_1.2,0.39225599765777586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,16,power_law_1.01,0.1573632001876831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,16,power_law_1.01,0.195904004573822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,16,1,power_law_1.2,0.47569918632507324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,16,power_law_1.2,0.1098688006401062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,16,1,power_law_1.2,0.6482816219329834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,16,power_law_1.01,0.1023743987083435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,16,power_law_1.01,0.10377600193023681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,16,1,power_law_1.2,0.8165311813354492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,16,power_law_1.01,0.10746879577636718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,16,1,balanced,0.13501333196957907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,16,1,balanced,0.1349120040734609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,16,power_law_1.01,0.11155200004577637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,16,1,balanced,0.13498133420944214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,16,1,balanced,0.13593066732088724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,16,1,power_law_1.2,1.1648896217346192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,16,power_law_1.01,0.1155135989189148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,16,1,balanced,0.134634663661321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,16,1,balanced,0.13803199927012125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,16,power_law_1.01,0.12455040216445923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,16,1,balanced,0.15842666228612265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,16,1,balanced,0.16038399934768677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,16,power_law_1.01,0.132697594165802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,16,1,power_law_1.01,1.446598434448242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,16,1,power_law_1.2,1.5124223709106446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,16,1,balanced,0.02517866591612498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,16,power_law_1.2,0.11286400556564331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,16,1,power_law_1.01,0.02202879935503006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,16,power_law_1.2,0.12424960136413574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,16,1,power_law_1.01,0.02512640058994293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,16,power_law_1.2,0.1319808006286621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,16,1,power_law_1.01,0.04333440065383911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,16,power_law_1.2,0.1404031991958618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,16,balanced,1.7771466573079426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,16,1,power_law_1.01,0.06256639957427979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,16,1,power_law_1.01,0.07528319954872131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,16,1,power_law_1.01,0.1117311954498291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,16,1,balanced,0.16476800044377646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,16,balanced,2.4022720654805503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,16,power_law_1.01,0.24466559886932374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,16,power_law_1.01,0.14313600063323975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,16,1,balanced,0.03159466634194056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,16,1,power_law_1.01,0.024774399399757386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,16,1,balanced,0.039701332648595176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,16,power_law_1.01,0.16193920373916626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,16,1,balanced,0.0613973339398702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,16,1,balanced,0.10110933581988017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,16,power_law_1.01,0.19987839460372925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,16,power_law_1.01,0.23915519714355468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,16,power_law_1.01,0.32614400386810305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,16,power_law_1.2,0.16335359811782837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,16,power_law_1.2,0.2052288055419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,16,power_law_1.01,0.42671999931335447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,16,power_law_1.2,0.2661119937896729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,16,1,power_law_1.01,0.11447039842605591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,16,power_law_1.2,0.34792959690093994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,16,1,power_law_1.01,0.11667840480804444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,16,power_law_1.01,0.3275775909423828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,16,power_law_1.2,0.4391295909881592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,16,1,power_law_1.01,0.030508801341056824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,16,1,balanced,0.12967466314633688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,16,power_law_1.01,0.4155007839202881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,16,1,balanced,0.1318666636943817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,16,1,power_law_1.01,0.04259839951992035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,16,1,balanced,0.13312000036239624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,16,1,balanced,0.13196266690889993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,16,1,balanced,0.13201600313186646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,16,1,balanced,0.13594667116800943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,16,power_law_1.2,0.06319360136985779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,16,1,balanced,0.14013333121935526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,16,1,balanced,0.1418826679388682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,16,power_law_1.2,0.08024320006370544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,16,power_law_1.2,0.09748479723930359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,16,power_law_1.01,0.6077439785003662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,16,power_law_1.01,0.7534272193908691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,16,power_law_1.2,0.6109055995941162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,16,power_law_1.01,1.065113639831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,16,power_law_1.01,0.5970176219940185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,16,1,balanced,0.14589333534240723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,16,power_law_1.01,0.7595327854156494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,16,1,balanced,0.1467519998550415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,16,power_law_1.01,1.391321563720703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,16,1,balanced,0.1511146624883016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,16,1,balanced,0.15828800201416016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,16,1,balanced,0.17083734273910522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,16,1,balanced,0.1830293337504069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,16,power_law_1.01,1.052454376220703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,16,1,balanced,0.2752693295478821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,16,1,balanced,0.30110400915145874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,16,power_law_1.2,0.8159423828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,16,1,power_law_1.01,0.1214143991470337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,16,1,balanced,0.42025601863861084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,16,power_law_1.01,2.167667198181152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,16,1,power_law_1.01,0.1249343991279602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,16,1,balanced,0.47413333257039386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,16,power_law_1.01,1.4382080078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,16,1,power_law_1.01,0.1363968014717102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,16,power_law_1.2,1.206220817565918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,16,1,balanced,0.7087039947509766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,16,1,power_law_1.01,0.06307200193405152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,16,1,power_law_1.01,0.15055999755859376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,16,1,balanced,0.8811733722686768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,16,1,power_law_1.01,0.15054080486297608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,16,1,power_law_1.01,0.15843839645385743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,16,power_law_1.01,2.715577507019043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,16,1,power_law_1.01,0.1619647979736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,16,power_law_1.2,0.0984063982963562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,16,1,power_law_1.01,0.16846079826354982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,16,1,power_law_1.01,0.17997440099716186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,16,1,power_law_1.01,0.20135040283203126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,16,1,power_law_1.01,0.22433919906616212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,16,1,power_law_1.01,0.27842559814453127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,16,1,power_law_1.01,0.07442560195922851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,16,1,power_law_1.01,0.3328511953353882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,16,1,power_law_1.01,0.10689280033111573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,16,power_law_1.01,2.112518310546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,16,1,balanced,1.2872107028961182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,16,1,power_law_1.01,0.11057920455932617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,16,1,power_law_1.01,0.4326591968536377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,16,1,balanced,0.16531733671824136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,16,1,balanced,0.1686720053354899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,16,power_law_1.2,0.10143359899520873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,16,1,balanced,1.6969920794169109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,16,1,power_law_1.01,0.5317440032958984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,16,power_law_1.2,0.10298880338668823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,16,1,power_law_1.2,0.026374399662017822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,16,power_law_1.2,0.10455039739608765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,16,1,power_law_1.2,0.028441599011421202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,16,power_law_1.01,2.740134429931641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,16,1,power_law_1.2,0.02184319943189621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,16,1,power_law_1.2,0.04309119880199432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,16,1,power_law_1.2,0.03277440071105957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,16,1,power_law_1.2,0.05950719714164734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,16,1,power_law_1.2,0.04289279878139496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,16,1,power_law_1.2,0.07291520237922669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,16,power_law_1.2,1.6030719757080079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,16,1,power_law_1.2,0.10488959550857543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,16,1,power_law_1.2,0.11356159448623657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,16,1,power_law_1.01,0.11391359567642212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,16,1,power_law_1.2,0.11498880386352539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,16,1,balanced,0.1742560068766276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,16,1,power_law_1.01,0.12076159715652465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,16,1,power_law_1.2,0.1218176007270813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,16,1,power_law_1.01,0.12383999824523925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,16,1,power_law_1.2,0.1257024049758911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,16,1,power_law_1.01,0.7383935928344727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,16,power_law_1.2,0.10760960578918458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,16,balanced,0.07273066540559132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,16,balanced,0.08867200215657552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,16,power_law_1.2,0.1125823974609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,16,balanced,0.12481600046157837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,16,balanced,0.2344320019086202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,16,power_law_1.2,0.11605119705200195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,16,balanced,0.3445386489232381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,16,balanced,0.4206666549046834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,16,balanced,0.6954933007558187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,16,balanced,0.704912026723226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,16,1,balanced,0.1850186586380005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,16,1,power_law_1.2,0.06059520244598389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,16,1,balanced,0.19803732633590698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,16,1,balanced,0.29020800193150836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,16,1,power_law_1.2,0.07406079769134521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,16,1,balanced,0.312389334042867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,16,1,power_law_1.2,0.13672959804534912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,16,1,power_law_1.2,0.10986239910125732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,16,1,power_law_1.2,0.1338495969772339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,16,1,power_law_1.2,0.11467519998550416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,16,1,power_law_1.01,0.9438655853271485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,16,1,power_law_1.2,0.13471360206604005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,16,1,power_law_1.2,0.14198399782180787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,16,power_law_1.2,0.12540160417556762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,16,1,power_law_1.2,0.14646400213241578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,16,1,power_law_1.01,1.3527551651000977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,16,1,power_law_1.2,0.15371520519256593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,16,balanced,0.7151093482971191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,16,balanced,0.7256960074106852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,16,1,power_law_1.01,1.75982723236084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,16,1,power_law_1.01,0.1301695942878723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,16,balanced,0.7450026671091715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,16,1,balanced,0.4400266806284587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,16,balanced,0.763429323832194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,16,1,balanced,0.4853760004043579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,16,1,power_law_1.2,0.11968640089035035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,16,1,balanced,0.7259199619293213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,16,1,power_law_1.2,0.12294399738311768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,16,1,balanced,0.892090638478597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,16,1,power_law_1.2,0.13047679662704467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,16,1,power_law_1.2,0.14088319540023803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,16,1,balanced,1.301194667816162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,16,1,power_law_1.2,0.16506880521774292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,16,1,power_law_1.2,0.19119999408721924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,16,1,power_law_1.2,0.21395840644836425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,16,balanced,0.04808000226815542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,16,balanced,0.0784853349129359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,16,balanced,0.1076693336168925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,16,balanced,0.16134933630625406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,16,balanced,0.2537813385327657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,16,balanced,0.7872746785481771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,16,balanced,0.32095466057459515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,16,balanced,0.8242239952087402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,16,balanced,0.6005333264668783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,16,power_law_1.2,0.13534719944000245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,16,balanced,0.6112533410390218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,16,balanced,0.8573386669158936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,16,balanced,0.6200053294499716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,16,balanced,0.8978239695231119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,16,balanced,0.6314773162206014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,16,balanced,0.9708960056304932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,16,balanced,0.651311993598938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,16,balanced,0.6689973672231039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,16,balanced,1.1136000156402588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,16,balanced,0.6946720282236735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,16,1,power_law_1.01,0.13164160251617432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,16,balanced,1.2585439682006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,16,1,power_law_1.01,0.1322816014289856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,16,1,power_law_1.01,0.13899519443511962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,16,1,power_law_1.01,0.14480639696121217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,16,1,power_law_1.01,0.1510208010673523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,16,power_law_1.2,0.14280320405960084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,16,1,power_law_1.01,0.16059520244598388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,16,power_law_1.2,0.1635648012161255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,16,1,power_law_1.01,0.18547840118408204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,16,1,balanced,1.7142346700032551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,16,power_law_1.2,0.20855679512023925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,16,1,power_law_1.01,0.21089279651641846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,16,power_law_1.2,0.25328640937805175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,16,1,power_law_1.2,0.15118080377578735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,16,1,power_law_1.01,0.26603519916534424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,16,power_law_1.2,2.309696006774902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,16,1,power_law_1.2,0.26924159526824953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,16,power_law_1.2,0.3379647970199585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,16,balanced,0.07132799923419952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,16,balanced,0.7309707005818685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,16,balanced,0.08540266752243042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,16,1,power_law_1.2,0.321126389503479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,16,power_law_1.2,0.4482560157775879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,16,balanced,0.7702346642812093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,16,1,power_law_1.2,0.4190336227416992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,16,balanced,1.6809013684590657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,16,balanced,0.8087680339813232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,16,power_law_1.2,0.5960959911346435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,16,1,power_law_1.2,0.5231808185577392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,16,balanced,0.8862186272939047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,16,power_law_1.2,0.8282815933227539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,16,balanced,1.0418933232625325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,16,1,power_law_1.2,0.729856014251709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,16,balanced,1.1981066862742107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,16,power_law_1.2,1.192448043823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,16,1,power_law_1.2,0.15379199981689454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,16,1,power_law_1.01,0.3169663906097412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,16,balanced,1.662559986114502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,16,balanced,0.11199466387430827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,16,balanced,0.18918933471043906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,16,1,power_law_1.01,0.41683201789855956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,16,balanced,0.326362669467926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,16,power_law_1.2,1.4714943885803222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,16,balanced,0.42258667945861816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,16,balanced,1.9681386947631836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,16,1,power_law_1.01,0.5206399917602539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,16,balanced,0.4313013156255086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,16,balanced,0.44302932421366376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,16,balanced,1.9733386039733887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,16,balanced,0.4511839946111043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,16,balanced,0.4605546792348226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,16,balanced,0.4803200165430705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,16,balanced,0.5001066525777181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,16,balanced,2.837594668070475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,16,balanced,0.5233920017878214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,16,1,power_law_1.2,0.16172159910202027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,16,balanced,0.5591839949289957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,16,1,power_law_1.2,0.1654207944869995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,16,1,power_law_1.2,0.9331392288208008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,16,1,power_law_1.2,0.17089279890060424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,16,balanced,3.5803521474202475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,16,1,power_law_1.01,0.7260735988616943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,16,power_law_1.2,3.1853376388549806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,16,balanced,2.7409226099650064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,16,1,power_law_1.01,0.9261631965637207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,16,power_law_1.01,0.2930943965911865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,16,1,power_law_1.01,1.339577579498291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,16,balanced,0.5940053462982178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,16,balanced,3.3617493311564126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,16,balanced,0.6356853246688843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,16,1,power_law_1.2,0.18192000389099122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,16,balanced,0.7112747033437093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,16,1,power_law_1.01,1.7526655197143555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,16,1,power_law_1.2,0.2054527997970581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,16,balanced,0.8543146451314291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,16,1,power_law_1.2,0.22899200916290283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,16,power_law_1.01,0.19400960206985474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,16,1,power_law_1.2,0.2830080032348633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,16,balanced,5.132800102233887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,16,1,power_law_1.2,0.3359935998916626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,16,power_law_1.01,0.2583679914474487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,16,balanced,5.302842775980632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,16,1,power_law_1.2,0.4377280235290527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,16,power_law_1.01,0.39516799449920653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,16,power_law_1.01,0.25462400913238525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,16,power_law_1.2,2.303264045715332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,16,power_law_1.01,0.3695807933807373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,16,power_law_1.01,0.25496959686279297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,16,power_law_1.01,0.2837440013885498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,16,power_law_1.01,0.39772160053253175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,16,balanced,1.0012959639231365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,16,balanced,6.765941619873047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,16,power_law_1.01,0.3067712068557739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,16,balanced,1.4158399899800618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,16,balanced,6.891183853149414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,16,power_law_1.01,0.5874815940856933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,16,balanced,1.7046559651692708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,16,1,power_law_1.2,0.5348735809326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,16,power_law_1.01,0.5792640209197998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,16,balanced,2.4072747230529785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,16,power_law_1.01,0.6060800075531005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,16,power_law_1.01,0.617580795288086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,16,power_law_1.01,0.39800319671630857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,16,balanced,2.975642522176107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,16,power_law_1.01,0.6374080181121826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,16,power_law_1.01,0.38453121185302735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,16,power_law_1.01,0.6599679946899414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,16,power_law_1.01,0.6531648159027099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,16,power_law_1.01,0.6876160144805908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,16,balanced,4.381909370422363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,16,power_law_1.01,0.6729599952697753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,16,power_law_1.01,0.718943977355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,16,power_law_1.01,0.7874495983123779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,16,balanced,10.164559682210287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,16,power_law_1.01,0.8392704010009766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,16,balanced,10.275093078613281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,16,balanced,5.668053309122722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,16,power_law_1.01,0.9471808433532715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,16,power_law_1.2,3.0846464157104494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,16,power_law_1.01,1.1422656059265137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,16,power_law_1.01,0.6535295963287353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,16,power_law_1.01,1.3467967987060547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,16,power_law_1.01,0.19734400510787964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,16,power_law_1.01,0.7001344203948975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,16,1,power_law_1.2,0.7409088134765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,16,power_law_1.01,0.25458559989929197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,16,power_law_1.01,1.7742912292480468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,16,balanced,8.384346644083658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,16,power_law_1.01,0.24685440063476563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,16,balanced,13.309125264485678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,16,power_law_1.01,0.2819200038909912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,16,power_law_1.01,2.1274431228637694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,16,balanced,13.452565511067709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,16,power_law_1.2,0.29292800426483157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,16,power_law_1.01,0.35469439029693606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,16,power_law_1.2,0.19477759599685668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,16,power_law_1.01,0.3893887996673584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,16,power_law_1.01,0.6982079982757569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,16,power_law_1.01,3.0600576400756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,16,power_law_1.2,0.23574399948120117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,16,power_law_1.01,0.41269121170043943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,16,power_law_1.01,0.7406400203704834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,16,power_law_1.2,0.24578559398651123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,16,power_law_1.01,0.42508158683776853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,16,power_law_1.01,0.7755775928497315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,16,power_law_1.01,0.43847041130065917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,16,power_law_1.2,0.26042881011962893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,16,power_law_1.01,0.815129566192627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,16,power_law_1.01,3.7828289031982423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,16,power_law_1.01,0.44582400321960447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,16,power_law_1.2,0.2600127935409546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,16,power_law_1.2,0.3722624063491821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,16,1,power_law_1.2,0.9492671966552735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,16,power_law_1.2,0.3067327976226807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,16,power_law_1.2,0.3803584098815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,16,power_law_1.2,0.5777599811553955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,16,power_law_1.2,0.3946943998336792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,16,balanced,10.950443267822266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,16,power_law_1.2,0.5922944068908691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,16,power_law_1.2,0.3978624105453491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,16,power_law_1.01,5.582182312011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,16,power_law_1.2,0.6055552005767822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,16,power_law_1.01,0.8483712196350097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,16,power_law_1.2,0.1972864031791687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,16,power_law_1.2,0.3937983989715576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,16,power_law_1.01,0.4685823917388916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,16,power_law_1.2,0.24184958934783934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,16,power_law_1.01,0.49590401649475097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,16,power_law_1.2,0.6352320194244385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,16,power_law_1.2,0.2447551965713501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,16,power_law_1.2,0.6502399921417237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,16,power_law_1.01,0.5187647819519043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,16,power_law_1.2,0.2850816011428833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,16,1,power_law_1.2,1.3486207962036132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,16,power_law_1.2,0.6834496021270752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,16,power_law_1.01,0.5523839950561523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,16,power_law_1.2,0.3218559980392456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,16,power_law_1.2,0.6855296134948731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,16,power_law_1.2,0.6178751945495605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,16,power_law_1.01,0.5935808181762695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,16,power_law_1.01,0.9033023834228515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,16,power_law_1.2,0.6913728237152099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,16,power_law_1.01,7.406771087646485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,16,power_law_1.2,0.6369984149932861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,16,power_law_1.01,0.6441472053527832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,16,power_law_1.2,0.7435520172119141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,16,power_law_1.2,0.663865613937378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,16,power_law_1.01,0.9897727966308594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,16,power_law_1.01,0.7371007919311523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,16,power_law_1.2,0.7716415882110595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,16,power_law_1.2,0.6855743885040283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,16,power_law_1.01,1.1714752197265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,16,power_law_1.01,0.9219264030456543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,16,power_law_1.2,0.8090368270874023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,16,power_law_1.01,1.0954303741455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,16,power_law_1.01,1.3768192291259767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,16,power_law_1.2,0.8549375534057617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,16,power_law_1.2,0.41254401206970215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,16,power_law_1.2,0.4006336212158203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,16,power_law_1.2,0.9124095916748047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,16,power_law_1.01,1.8016639709472657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,16,power_law_1.2,0.4210815906524658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,16,power_law_1.2,1.0291199684143066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,16,power_law_1.2,0.423360013961792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,16,power_law_1.2,0.745145606994629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,16,power_law_1.01,2.1673023223876955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,16,power_law_1.2,0.4375296115875244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,16,power_law_1.01,10.566957092285156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,16,power_law_1.2,1.2107904434204102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,16,power_law_1.2,0.7959487915039063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,16,1,power_law_1.2,1.7635711669921874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,16,power_law_1.2,0.4636223793029785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,16,1,balanced,0.06553600231806438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,16,1,balanced,0.09371733665466309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,16,1,balanced,0.14995200435320535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,16,power_law_1.2,0.49086718559265136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,16,power_law_1.2,0.8458623886108398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,16,power_law_1.01,1.480140781402588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,16,power_law_1.01,2.914246368408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,16,power_law_1.2,0.5208703994750976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,16,power_law_1.2,0.9670720100402832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,16,power_law_1.2,0.5666175842285156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,16,power_law_1.01,1.874687957763672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,16,power_law_1.2,0.6105663776397705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,16,power_law_1.2,1.1668607711791992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,16,power_law_1.2,0.6504960060119629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,16,power_law_1.01,3.779033660888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,16,power_law_1.01,2.5797760009765627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,16,power_law_1.2,1.3768511772155763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,16,1,balanced,0.3277226686477661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,16,1,balanced,0.5949493249257406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,16,power_law_1.2,0.7518911838531495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,16,1,balanced,0.8590559959411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,16,power_law_1.2,1.7935487747192382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,16,power_law_1.2,0.9478655815124511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,16,power_law_1.01,3.38155517578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,16,power_law_1.2,1.1302271842956544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,16,power_law_1.2,2.233158493041992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,16,power_law_1.01,5.456774520874023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,16,1,power_law_1.2,1.3591360092163085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,16,power_law_1.01,13.882412719726563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,16,power_law_1.2,1.4125887870788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,16,1,balanced,0.06377600133419037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,16,power_law_1.2,1.5008576393127442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,16,1,balanced,0.08620267113049825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,16,1,balanced,0.1379680037498474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,16,1,balanced,0.29502934217453003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,16,1,balanced,0.5301173528035482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,16,1,balanced,0.8624373277028402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,16,power_law_1.2,3.155564880371094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,16,1,balanced,0.7696266969045004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,16,1,balanced,0.8647786776224772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,16,power_law_1.01,4.882124710083008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,16,1,balanced,0.8702560265858968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,16,1,balanced,0.8720106283823649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,16,1,balanced,0.8794826666514078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,16,1,balanced,0.8891092936197916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,16,power_law_1.01,7.152178955078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,16,power_law_1.2,3.9616832733154297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,16,1,balanced,0.8994666735331217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,16,1,balanced,0.9129066467285156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,16,power_law_1.2,1.8836223602294921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,16,1,balanced,0.925269365310669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,16,1,balanced,0.774394671122233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,16,1,power_law_1.2,1.7699392318725586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,16,power_law_1.2,1.8096256256103516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,16,1,balanced,0.7781866391499838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,16,1,balanced,1.2565226554870605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,16,power_law_1.01,6.421657562255859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,16,1,balanced,0.7851946353912354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,16,1,power_law_1.01,0.059673601388931276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,16,1,balanced,0.7898346583048502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,16,1,balanced,0.7978986899058024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,16,power_law_1.2,2.7062015533447266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,16,1,power_law_1.01,0.09248639941215515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,16,power_law_1.2,2.3885120391845702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,16,1,balanced,0.808784008026123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,16,power_law_1.2,5.687488174438476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,16,1,power_law_1.01,0.14800000190734863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,16,1,balanced,0.8231840133666992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,16,1,balanced,0.8403200308481852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,16,1,power_law_1.01,0.2800447940826416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,16,1,balanced,0.8579733371734619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,16,power_law_1.2,3.157548713684082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,16,1,power_law_1.01,0.40506877899169924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,16,1,balanced,1.2005706628163655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,16,1,balanced,1.6272800763448079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,16,power_law_1.01,10.66776351928711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,16,1,balanced,1.5684800148010254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,16,1,power_law_1.01,0.6559807777404785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,16,1,balanced,1.130832036336263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,16,1,balanced,1.0197707017262776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,16,1,balanced,1.189296007156372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,16,1,power_law_1.01,0.6934720039367676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,16,1,balanced,1.084181308746338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,16,power_law_1.01,9.288921356201172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,16,power_law_1.2,4.064960098266601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,16,power_law_1.2,7.72200927734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,16,1,power_law_1.01,0.7208767890930176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,16,1,balanced,1.788464069366455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,16,power_law_1.2,3.4301246643066405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,16,1,power_law_1.01,0.7396992206573486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,16,1,balanced,1.9104320208231609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,16,1,power_law_1.01,0.7641407966613769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,16,1,balanced,2.721226692199707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,16,1,power_law_1.01,0.8356863975524902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,16,1,balanced,1.9443039894104004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,16,1,power_law_1.01,0.8867327690124511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,16,power_law_1.2,6.186726379394531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,16,power_law_1.2,5.054982376098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,16,1,balanced,2.0572427113850913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,16,1,balanced,2.91048526763916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,16,power_law_1.01,14.487571716308594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,16,power_law_1.01,12.248799896240234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,16,1,balanced,2.9747947057088218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,16,1,balanced,3.158053398132324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,16,1,power_law_1.2,0.06115840077400207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,16,1,power_law_1.01,0.94900484085083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,16,power_law_1.2,6.596985626220703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,16,1,power_law_1.2,0.09279999732971192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,16,power_law_1.2,7.690239715576172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,16,1,power_law_1.01,1.0782527923583984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,16,1,power_law_1.2,0.14742399454116822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,16,1,balanced,4.591845194498698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,16,1,balanced,4.856927871704102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,16,1,power_law_1.2,0.2598655939102173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,16,1,power_law_1.01,0.049420800805091855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,16,1,power_law_1.2,0.3700608015060425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,16,1,power_law_1.01,0.08498560190200806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,16,1,power_law_1.2,0.6395391941070556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,16,1,power_law_1.01,0.13625600337982177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,16,1,power_law_1.2,0.6912831783294677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,16,1,balanced,5.949520111083984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,16,1,power_law_1.01,0.2637631893157959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,16,1,power_law_1.2,0.7239871978759765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,16,1,power_law_1.01,1.194803237915039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,16,power_law_1.2,9.642342376708985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,16,1,power_law_1.01,0.36360960006713866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,16,power_law_1.2,11.716563415527343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,16,1,power_law_1.01,1.3319168090820312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,16,power_law_1.2,12.581779479980469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,16,1,power_law_1.01,0.5812863826751709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,16,1,balanced,5.671205520629883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,16,1,power_law_1.01,0.630457592010498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,16,1,power_law_1.01,1.5989888191223145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,16,1,balanced,8.68222427368164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,16,1,power_law_1.01,0.6586304187774659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,16,1,power_law_1.01,1.2702207565307617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,16,1,power_law_1.01,0.6760960102081299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,16,1,power_law_1.2,0.7559872150421143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,16,1,power_law_1.01,1.4213503837585448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,16,1,power_law_1.01,0.6967616081237793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,16,1,power_law_1.2,0.7803647994995118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,16,1,power_law_1.01,0.7671103954315186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,16,1,balanced,8.385509490966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,16,1,power_law_1.01,1.7421184539794923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,16,1,power_law_1.2,0.8485119819641114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,16,1,power_law_1.01,0.8297280311584473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,16,1,power_law_1.2,0.9210559844970703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,16,1,power_law_1.01,2.0615423202514647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,16,1,power_law_1.01,0.8963775634765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,16,1,balanced,10.827434539794922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,16,1,power_law_1.2,0.9767423629760742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,16,1,power_law_1.01,1.0330047607421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,16,1,power_law_1.2,0.048793599009513855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,16,1,power_law_1.2,1.1140543937683105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,16,1,power_law_1.01,2.6972671508789063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,16,1,power_law_1.01,1.1415743827819824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,16,1,power_law_1.2,1.2508223533630372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,16,1,power_law_1.01,1.2827903747558593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,16,1,balanced,10.530869166056315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,32,1,balanced,0.02526933451493581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,32,1,balanced,0.029418667157491047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,32,1,balanced,0.0417546679576238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,32,1,balanced,0.04211199780305227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,32,1,balanced,0.0662613312403361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,16,1,power_law_1.2,1.3712639808654785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,16,1,power_law_1.01,3.3242496490478515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,16,1,power_law_1.01,1.5414912223815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,16,1,power_law_1.01,1.1486720085144042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,16,1,power_law_1.2,0.0864960014820099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,16,1,power_law_1.2,1.6392704010009767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,16,1,power_law_1.2,0.13539199829101561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,16,1,power_law_1.01,1.3163007736206054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,16,1,power_law_1.2,1.2949055671691894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,16,1,power_law_1.2,0.2478015899658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,16,1,power_law_1.01,4.606623840332031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,32,1,balanced,0.06470400094985962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,16,1,power_law_1.01,1.626246452331543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,32,1,balanced,0.07214933137098949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,32,1,balanced,0.07455466687679291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,32,1,balanced,0.06498666604359944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,16,1,power_law_1.2,1.4637375831604005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,32,1,balanced,0.07452266911665599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,32,1,balanced,0.07537066439787547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,32,1,balanced,0.07901333272457123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,32,1,balanced,0.07935466865698497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,32,1,balanced,0.0844586690266927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,32,1,balanced,0.09491200248400371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,32,1,balanced,0.09670399626096089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,16,1,power_law_1.01,1.9494335174560546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,16,1,power_law_1.2,1.7805120468139648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,16,power_law_1.2,12.72775650024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,16,1,power_law_1.01,5.876607894897461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,16,1,power_law_1.2,2.092473602294922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,16,1,power_law_1.2,0.3352960109710693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,16,1,power_law_1.2,0.5765696048736573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,32,1,balanced,0.09630399942398071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,32,1,balanced,0.10828800002733867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,32,1,balanced,0.11821867028872173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,32,1,balanced,0.1399893363316854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,32,1,balanced,0.17274133364359537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,32,1,balanced,0.24103999137878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,16,power_law_1.2,17.784288024902345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,32,1,balanced,0.2864266633987427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,32,1,balanced,0.41125865777333576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,32,1,balanced,0.02920000006755193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,32,1,balanced,0.5318613449732462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,32,1,balanced,0.7717920144399008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,32,1,balanced,1.0140533447265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,32,1,power_law_1.01,0.02707200050354004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,16,1,power_law_1.2,2.7176128387451173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,16,1,power_law_1.2,0.6333055973052979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,16,1,power_law_1.2,0.6470592021942139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,16,1,power_law_1.01,2.570675277709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,32,1,balanced,0.033520000676314034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,32,1,balanced,0.03563733398914337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,16,1,power_law_1.2,0.6843584060668946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,16,1,power_law_1.2,3.350297546386719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,16,1,power_law_1.2,0.7206463813781738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,16,1,power_law_1.01,8.483980560302735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,32,1,power_law_1.01,0.03918719887733459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,16,1,power_law_1.2,0.7863423824310303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,32,1,power_law_1.01,0.042828801274299624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,16,1,power_law_1.01,3.1964351654052736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,32,1,power_law_1.01,0.042131200432777405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,32,1,power_law_1.01,0.06423680186271667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,16,1,power_law_1.2,0.8577216148376465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,32,1,power_law_1.01,0.06431999802589417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,32,1,power_law_1.01,0.07085440158843995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,16,1,power_law_1.2,4.619385528564453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,32,1,power_law_1.01,0.07186560034751892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,16,1,power_law_1.2,0.9367808341979981
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,32,1,balanced,0.025573333104451496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,32,1,power_law_1.01,0.06563199758529663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,32,1,balanced,0.027285332481066387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,32,1,balanced,0.027632000545660656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,32,1,power_law_1.01,0.0762112021446228
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,32,1,balanced,0.02784000088771184
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,32,1,balanced,0.02757866680622101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,32,1,power_law_1.01,0.0801472008228302
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,32,1,balanced,0.02757333219051361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,16,1,power_law_1.2,1.0590656280517579
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,32,1,balanced,0.02959999938805898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,32,1,power_law_1.01,0.08376320004463196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,16,1,power_law_1.01,4.4699455261230465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,32,1,power_law_1.01,0.08511999845504761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,32,1,power_law_1.01,0.08840320110321045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,16,1,power_law_1.2,1.1968255996704102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,32,1,power_law_1.01,0.09840000271797181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,16,1,power_law_1.2,5.89997444152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,16,1,power_law_1.2,1.3308352470397948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,32,1,balanced,0.037717332442601524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,32,1,balanced,0.05163733164469401
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,32,1,balanced,0.051674668987592064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,32,1,power_law_1.01,0.10275839567184449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,32,1,balanced,0.05202133456865946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,32,1,balanced,0.05203733344872793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,32,1,balanced,0.05197333296140035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,16,1,power_law_1.01,11.056998443603515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,32,1,balanced,0.054117331902186074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,32,1,power_law_1.01,0.0983295977115631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,32,1,balanced,0.05611733098824819
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,32,1,balanced,0.05589866638183594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,32,1,balanced,0.029696000119050343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,32,1,balanced,0.05811200042565664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,32,1,balanced,0.03141866624355316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,32,1,balanced,0.0609440008799235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,32,1,balanced,0.031498665610949196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,32,1,balanced,0.06935999790827434
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,32,1,balanced,0.03359466542800268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,32,1,balanced,0.07335466643174489
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,32,1,balanced,0.035589332381884255
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,32,1,balanced,0.035546667873859406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,32,1,balanced,0.03770666569471359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,32,1,balanced,0.044346665342648826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,32,1,balanced,0.04373333354791006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,32,1,balanced,0.04313066601753235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,32,1,balanced,0.060362666845321655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,32,1,balanced,0.05643733342488607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,32,1,balanced,0.07348800202210744
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,32,1,balanced,0.08690667152404785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,32,1,balanced,0.11115200320879619
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,32,1,balanced,0.13808533549308777
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,32,1,balanced,0.19534399112065634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,32,1,balanced,0.25
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,32,1,balanced,0.3615093231201172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,32,1,power_law_1.01,0.028115200996398925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,16,1,power_law_1.01,5.739923095703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,32,1,balanced,0.46591468652089435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,32,1,balanced,0.07447466750939687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,32,1,balanced,0.08619733651479085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,32,1,balanced,0.10269332925478618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,32,1,balanced,0.13428800304730734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,32,1,balanced,0.17333867152531943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,16,1,power_law_1.2,8.508716583251953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,32,1,balanced,0.23013333479563394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,32,1,balanced,0.2923520008722941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,32,1,balanced,0.4154026508331299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,32,1,balanced,0.5400799910227457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,32,1,balanced,0.7899359862009684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,16,1,power_law_1.2,1.584620761871338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,32,1,balanced,1.0402613480885823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,32,1,power_law_1.01,0.03374080061912536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,14336,2,8,32,1,power_law_1.2,0.0240447998046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,14336,2,8,32,1,power_law_1.2,0.039603200554847715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,16,1,power_law_1.2,1.1846528053283691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,14336,2,8,32,1,power_law_1.2,0.041356799006462094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,32,1,power_law_1.01,0.024140800535678863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,14336,2,8,32,1,power_law_1.2,0.04421760141849518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,16,1,power_law_1.01,8.344198608398438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,32,1,power_law_1.01,0.02460159957408905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,14336,2,8,32,1,power_law_1.2,0.07040640115737914
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,32,1,power_law_1.01,0.02540160119533539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,16,1,power_law_1.2,1.348646354675293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,32,1,power_law_1.01,0.026073598861694337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,32,1,power_law_1.01,0.02645759880542755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,32,1,power_law_1.01,0.02743679881095886
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,32,1,power_law_1.01,0.02824319899082184
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,32,1,power_law_1.01,0.02975359857082367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,32,1,power_law_1.01,0.034668800234794614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,32,1,power_law_1.01,0.029862400889396668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,32,1,power_law_1.01,0.03648000061511993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,16,1,power_law_1.2,11.089337921142578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,32,1,power_law_1.01,0.030188798904418945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,14336,2,8,32,1,power_law_1.2,0.027673599123954774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,32,1,power_law_1.01,0.032979199290275575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,14336,2,8,32,1,power_law_1.2,0.03307519853115082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,32,1,power_law_1.01,0.03378559947013855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,14336,2,8,32,1,power_law_1.2,0.07364479899406433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,32,1,power_law_1.01,0.03468160033226013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,14336,2,8,32,1,power_law_1.2,0.07448959946632386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,14336,2,8,32,1,power_law_1.2,0.07657600045204163
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,32,1,power_law_1.01,0.03797760009765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,14336,2,8,32,1,power_law_1.2,0.07517439723014832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,14336,2,8,32,1,power_law_1.2,0.07982720136642456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,14336,2,8,32,1,power_law_1.2,0.08229119777679443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,32,1,power_law_1.01,0.039001598954200745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,32,1,power_law_1.01,0.04046080112457275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,32,1,power_law_1.01,0.04986239969730377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,16,1,power_law_1.01,10.899533081054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,32,1,power_law_1.01,0.04349440038204193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,32,1,power_law_1.01,0.1118016004562378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,16,power_law_1.2,15.447001647949218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,32,1,balanced,0.03345600018898646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,32,1,balanced,0.060229331254959106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,32,1,power_law_1.01,0.12139519453048705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,32,1,power_law_1.01,0.053273600339889524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,16,1,power_law_1.2,1.6629119873046876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,14336,2,8,32,1,power_law_1.2,0.0833791971206665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,16,1,power_law_1.2,1.9718463897705079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,14336,2,8,32,1,power_law_1.2,0.08716800212860107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,14336,2,8,32,1,power_law_1.2,0.03375360071659088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,32,1,balanced,0.07087466617425282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,32,1,balanced,0.0711359977722168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,32,1,power_law_1.01,0.14819200038909913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,14336,2,8,32,1,power_law_1.2,0.09216639995574952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,32,1,balanced,0.111653337876002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,32,1,balanced,0.11329066753387451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,14336,2,8,32,1,power_law_1.2,0.0985152006149292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,16,1,power_law_1.2,2.603596878051758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,14336,2,8,32,1,power_law_1.2,0.10266239643096924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,14336,2,8,32,1,power_law_1.2,0.09809280037879944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,32,1,power_law_1.01,0.05006080269813538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,32,1,power_law_1.01,0.05178239941596985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,16,1,power_law_1.2,3.2248191833496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,32,1,power_law_1.01,0.052243202924728394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,14336,2,8,32,1,power_law_1.2,0.036595198512077334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,32,1,power_law_1.01,0.05374079942703247
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,14336,2,8,32,1,power_law_1.2,0.02561280131340027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,14336,2,8,32,1,power_law_1.2,0.04967679977416992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,32,1,power_law_1.01,0.054099202156066895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,14336,2,8,32,1,power_law_1.2,0.02535040080547333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,14336,2,8,32,1,power_law_1.2,0.05021439790725708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,32,1,power_law_1.01,0.057683199644088745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,14336,2,8,32,1,power_law_1.2,0.026598399877548216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,32,1,power_law_1.01,0.17252479791641234
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,14336,2,8,32,1,power_law_1.2,0.02602880001068115
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,14336,2,8,32,1,power_law_1.2,0.026662400364875792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,14336,2,8,32,1,power_law_1.2,0.02766079902648926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,14336,2,8,32,1,power_law_1.2,0.028729599714279175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,32,1,power_law_1.01,0.06064640283584595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,14336,2,8,32,1,power_law_1.2,0.11088639497756958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,14336,2,8,32,1,power_law_1.2,0.030246400833129884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,14336,2,8,32,1,power_law_1.2,0.030726400017738343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,14336,2,8,32,1,power_law_1.2,0.03091840147972107
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,14336,2,8,32,1,power_law_1.2,0.03265919983386993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,32,1,balanced,0.11507733662923177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,32,1,balanced,0.11595199505488078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,32,1,balanced,0.11538133025169373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,14336,2,8,32,1,power_law_1.2,0.03310079872608185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,32,1,balanced,0.11775466799736023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,32,1,balanced,0.11952533324559529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,32,1,balanced,0.12180800239245097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,32,1,balanced,0.12187199791272481
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,32,1,power_law_1.01,0.07184000015258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,14336,2,8,32,1,power_law_1.2,0.12040319442749023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,16,1,power_law_1.2,4.500390243530274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,14336,2,8,32,1,power_law_1.2,0.05122560262680054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,32,1,power_law_1.01,0.059545600414276124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,14336,2,8,32,1,power_law_1.2,0.05221760272979736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,32,1,power_law_1.01,0.06060799956321716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,32,1,balanced,0.12843733032544455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,32,1,balanced,0.14103999733924866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,32,1,power_law_1.01,0.0636672019958496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,32,1,balanced,0.14203733205795288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,32,1,balanced,0.13768532872200012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,32,1,power_law_1.01,0.0689087986946106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,32,1,balanced,0.1520853340625763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,16,1,power_law_1.2,5.768307113647461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,32,1,balanced,0.16685332854588827
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,32,1,power_law_1.01,0.08734080195426941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,32,1,balanced,0.19945067167282104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,32,1,balanced,0.25629866123199463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,32,1,balanced,0.3558773199717204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,32,1,balanced,0.44284268220265705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,32,1,balanced,0.6404159863789877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,32,1,balanced,0.8372853597005209
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,14336,2,8,32,1,power_law_1.2,0.03450239896774292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,32,1,balanced,1.2293972969055176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,14336,2,8,32,1,power_law_1.2,0.15094399452209473
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,14336,2,8,32,1,power_law_1.2,0.03701759874820709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,32,1,balanced,1.6218454043070476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,32,1,balanced,0.03340800106525421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,32,1,balanced,0.04062933226426443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,14336,2,8,32,1,power_law_1.2,0.03859840035438537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,32,1,power_law_1.01,0.25100159645080566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,32,1,power_law_1.01,0.07642880082130432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,32,1,power_law_1.01,0.07916160225868225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,14336,2,8,32,1,power_law_1.2,0.05539839863777161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,32,1,power_law_1.01,0.097324800491333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,14336,2,8,32,1,power_law_1.2,0.03948160111904144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,14336,2,8,32,1,power_law_1.2,0.056377601623535153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,14336,2,8,32,1,power_law_1.2,0.057657599449157715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,16,1,power_law_1.2,8.381721496582031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,14336,2,8,32,1,power_law_1.2,0.060838401317596436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,14336,2,8,32,1,power_law_1.2,0.06244480013847351
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,32,1,power_law_1.01,0.11128319501876831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,14336,2,8,32,1,power_law_1.2,0.06639360189437866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,14336,2,8,32,1,power_law_1.2,0.04213759899139404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,32,1,power_law_1.01,0.10929280519485474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,14336,2,8,32,1,power_law_1.2,0.05361279845237732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,32,1,power_law_1.01,0.149619197845459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,32,1,balanced,0.053743998209635414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,32,1,balanced,0.054010664423306785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,32,1,balanced,0.07056533296902974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,32,1,balanced,0.07141866783301036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,32,1,balanced,0.07234133283297221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,32,1,balanced,0.08365333080291748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,32,1,balanced,0.0743999977906545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,32,1,balanced,0.07491733133792877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,32,1,balanced,0.0811359981695811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,32,1,balanced,0.08695466319719951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,32,1,power_law_1.01,0.17851519584655762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,14336,2,8,32,1,power_law_1.2,0.0714303970336914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,14336,2,8,32,1,power_law_1.2,0.07667840123176575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,14336,2,8,32,1,power_law_1.2,0.17466239929199218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,14336,2,8,32,1,power_law_1.2,0.07996799945831298
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,14336,2,8,32,1,power_law_1.2,0.060550397634506224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,32,1,balanced,0.0897706647713979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,32,1,balanced,0.09290666381518047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,32,1,balanced,0.10883200168609619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,32,1,balanced,0.11562666296958923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,32,1,balanced,0.11133866508801778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,32,1,balanced,0.12829867005348206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,32,1,balanced,0.14282133181889853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,32,1,balanced,0.2035306692123413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,32,1,balanced,0.2450773318608602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,32,1,balanced,0.34195733070373535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,14336,2,8,32,1,power_law_1.2,0.0965503990650177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,32,1,balanced,0.43646931648254395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,32,1,balanced,0.6347999970118204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,32,1,balanced,0.8350346883138021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,32,1,balanced,1.2310453255971272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,32,1,power_law_1.01,0.30369279384613035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,32,1,power_law_1.01,0.13776639699935914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,32,1,balanced,1.6272800763448079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,14336,2,8,32,1,power_law_1.2,0.1074112057685852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,32,1,balanced,0.04633066554864248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,32,1,balanced,0.049866666396458946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,32,1,balanced,0.06218666831652323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,32,1,balanced,0.0621973325808843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,32,1,balanced,0.06354133288065593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,14336,2,8,32,1,power_law_1.2,0.14643839597702027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,32,1,balanced,0.06292800108591716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,32,1,balanced,0.06406400104363759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,32,1,balanced,0.06418133278687795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,32,1,balanced,0.06423999865849812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,14336,2,8,32,1,power_law_1.2,0.07256320118904114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,16,1,power_law_1.2,10.971353912353516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,32,1,power_law_1.01,0.2425152063369751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,32,1,balanced,0.06438399851322174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,32,1,balanced,0.06687999765078227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,14336,2,8,32,1,power_law_1.2,0.17468160390853882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,32,1,balanced,0.07034666836261749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,32,1,balanced,0.07236266632874806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,32,1,balanced,0.08004266520341237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,32,1,balanced,0.0881760021050771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,32,1,balanced,0.10225600004196167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,32,1,balanced,0.09835199515024821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,32,1,balanced,0.11755733688672383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,32,1,balanced,0.13116799791653952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,32,1,balanced,0.19662932554880777
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,14336,2,8,32,1,power_law_1.2,0.08592640161514283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,32,1,balanced,0.027450665831565857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,32,1,balanced,0.029311999678611755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,32,1,balanced,0.029631999631722767
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,32,1,balanced,0.029215998947620392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,32,1,balanced,0.02958400050799052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,32,1,balanced,0.029482667644818623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,32,1,balanced,0.03334933271010717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,32,1,balanced,0.033589333295822144
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,32,1,balanced,0.03433600068092346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,32,1,balanced,0.03534399966398875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,32,1,balanced,0.03580799947182337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,32,1,balanced,0.03940266619126002
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,32,1,balanced,0.039546666045983635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,32,1,balanced,0.04163199911514918
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,32,1,balanced,0.05006400247414907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,14336,2,8,32,1,power_law_1.2,0.2475584030151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,32,1,balanced,0.05004799862702688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,32,1,power_law_1.01,0.30543999671936034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,32,1,balanced,0.049135997891426086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,32,1,balanced,0.07277333239714305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,32,1,balanced,0.07445333401362102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,32,1,balanced,0.09274133046468098
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,32,1,balanced,0.10930666327476501
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,32,1,balanced,0.1486026644706726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,32,1,balanced,0.18981866041819254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,32,1,balanced,0.2716053326924642
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,32,1,balanced,0.346890648206075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,32,1,balanced,0.5040533145268759
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,32,1,balanced,0.6544959942499796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,32,1,power_law_1.01,0.050809597969055174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,32,1,balanced,0.2304906646410624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,32,1,power_law_1.01,0.05438079833984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,32,1,power_law_1.01,0.1946943998336792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,32,1,power_law_1.01,0.4268352031707764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,14336,2,8,32,1,power_law_1.2,0.11080960035324097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,32,1,balanced,0.323855996131897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,32,1,power_law_1.01,0.05649920105934143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,32,1,balanced,0.42089064915974933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,32,1,power_law_1.01,0.06965759992599488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,32,1,balanced,0.5982880194981893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,32,1,power_law_1.01,0.11030399799346924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,32,1,balanced,0.7772906621297201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,32,1,power_law_1.01,0.11119359731674194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,32,1,power_law_1.01,0.11214720010757447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,32,1,balanced,1.1409813563028972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,32,1,power_law_1.01,0.11317119598388672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,32,1,power_law_1.01,0.11562880277633666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,32,1,balanced,1.5005812644958496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,32,1,power_law_1.01,0.1171455979347229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,32,1,power_law_1.01,0.03537279963493347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,32,1,power_law_1.01,0.11947519779205322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,32,1,power_law_1.01,0.03999359905719757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,32,1,power_law_1.01,0.12485760450363159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,32,1,power_law_1.01,0.12704000473022461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,32,1,power_law_1.01,0.13147519826889037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,32,1,power_law_1.01,0.13831679821014403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,32,1,power_law_1.01,0.1440832018852234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,32,1,power_law_1.01,0.1420799970626831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,32,1,power_law_1.01,0.05663999915122986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,14336,2,8,32,1,power_law_1.2,0.24131200313568116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,32,1,power_law_1.01,0.05752320289611816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,32,1,power_law_1.01,0.15980160236358643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,32,1,power_law_1.01,0.06836479902267456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,32,1,power_law_1.01,0.07052800059318542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,32,1,power_law_1.01,0.07493759989738465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,32,1,power_law_1.01,0.17562240362167358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,32,1,power_law_1.01,0.4268223762512207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,32,1,power_law_1.01,0.22648959159851073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,14336,2,8,32,1,power_law_1.2,0.30380799770355227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,32,1,power_law_1.01,0.07275519967079162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,32,1,power_law_1.01,0.08261759877204895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,32,1,power_law_1.01,0.07637119889259339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,32,1,power_law_1.01,0.08664960265159607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,32,1,power_law_1.01,0.09076480269432068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,32,1,power_law_1.01,0.09594240188598632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,32,1,power_law_1.01,0.10279680490493774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,32,1,power_law_1.01,0.281113600730896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,32,1,power_law_1.01,0.11071360111236572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,32,1,power_law_1.01,0.11634559631347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,32,1,power_law_1.01,0.37521278858184814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,32,1,power_law_1.01,0.1155776023864746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,14336,2,8,32,1,power_law_1.2,0.13783040046691894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,32,1,power_law_1.01,0.13749760389328003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,32,1,power_law_1.01,0.162336003780365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,32,1,power_law_1.01,0.49217920303344725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,32,1,power_law_1.01,0.5570047855377197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,32,1,power_law_1.01,0.5493887901306153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,14336,2,8,32,1,power_law_1.2,0.3023360013961792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,32,1,power_law_1.01,0.21422080993652343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,32,1,power_law_1.01,0.2615808010101318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,32,1,power_law_1.01,0.3666752099990845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,14336,2,8,32,1,power_law_1.2,0.19471360445022584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,14336,2,8,32,1,power_law_1.2,0.41623678207397463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,32,1,power_law_1.01,0.2525376081466675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,14336,2,8,32,1,power_law_1.2,0.4252927780151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,32,1,power_law_1.01,0.4625279903411865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,32,1,power_law_1.01,0.6626815795898438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,32,1,power_law_1.01,0.7950079917907715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,32,1,power_law_1.01,0.36014719009399415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,14336,2,8,32,1,power_law_1.2,0.5476160049438477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,32,1,power_law_1.01,0.8099136352539062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,32,1,power_law_1.01,0.8700223922729492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,32,1,power_law_1.01,0.6590976238250732
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,14336,2,8,32,1,power_law_1.2,0.2543936014175415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,32,1,power_law_1.01,1.2621312141418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,32,1,power_law_1.01,1.044262409210205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,32,1,power_law_1.01,0.04467200040817261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,32,1,power_law_1.01,0.05023360252380371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,32,1,power_law_1.01,0.04953599870204926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,32,1,power_law_1.01,0.05854719877243042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,32,1,power_law_1.01,0.06035199761390686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,32,1,power_law_1.01,0.06032000184059143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,32,1,power_law_1.01,0.06308479905128479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,32,1,power_law_1.01,0.06389120221138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,32,1,power_law_1.01,0.06339200139045716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,32,1,power_law_1.01,0.06389759778976441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,32,1,power_law_1.01,0.06704639792442321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,32,1,power_law_1.01,0.06930559873580933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,32,1,power_law_1.01,0.06945279836654664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,32,1,power_law_1.01,0.07812479734420777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,32,1,power_law_1.01,0.08635519742965699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,32,1,power_law_1.01,0.09950720071792603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,32,1,power_law_1.01,0.0987775981426239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,32,1,power_law_1.01,0.11639679670333862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,32,1,power_law_1.01,0.136735999584198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,32,1,power_law_1.01,0.8555904388427734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,14336,2,8,32,1,power_law_1.2,0.5508416175842286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,32,1,power_law_1.01,0.1945088028907776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,32,1,power_law_1.01,0.2299583911895752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,32,1,power_law_1.01,1.6580415725708009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,32,1,power_law_1.01,1.06428804397583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,32,1,power_law_1.01,0.3245568037033081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,16384,2,8,32,1,power_law_1.2,0.0511680006980896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,16384,2,8,32,1,power_law_1.2,0.05934079885482788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,16384,2,8,32,1,power_law_1.2,0.060236799716949466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,16384,2,8,32,1,power_law_1.2,0.06910079717636108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,16384,2,8,32,1,power_law_1.2,0.11055999994277954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,16384,2,8,32,1,power_law_1.2,0.11155200004577637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,16384,2,8,32,1,power_law_1.2,0.11264640092849731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,16384,2,8,32,1,power_law_1.2,0.11404160261154175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,16384,2,8,32,1,power_law_1.2,0.11556479930877686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,16384,2,8,32,1,power_law_1.2,0.11598720550537109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,16384,2,8,32,1,power_law_1.2,0.12009600400924683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,16384,2,8,32,1,power_law_1.2,0.12567039728164672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,16384,2,8,32,1,power_law_1.2,0.1280959963798523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,16384,2,8,32,1,power_law_1.2,0.13114880323410033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,16384,2,8,32,1,power_law_1.2,0.13879679441452025
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,32,1,power_law_1.01,0.026310399174690247
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,32,1,power_law_1.01,0.027142399549484254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,32,1,power_law_1.01,0.026867198944091796
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,32,1,power_law_1.01,0.027647998929023743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,16384,2,8,32,1,power_law_1.2,0.14358400106430053
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,32,1,power_law_1.01,0.027904000878334046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,32,1,power_law_1.01,0.029446399211883544
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,32,1,power_law_1.01,0.03129599988460541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,32,1,power_law_1.01,0.03498240113258362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,16384,2,8,32,1,power_law_1.2,0.14302719831466676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,32,1,power_law_1.01,0.03287039995193482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,16384,2,8,32,1,power_law_1.2,0.1593727946281433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,32,1,power_law_1.01,0.03406080007553101
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,32,1,power_law_1.01,0.037452799081802365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,32,1,power_law_1.01,0.037747201323509214
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,32,1,power_law_1.01,0.03806079924106598
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,32,1,power_law_1.01,0.04424319863319397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,32,1,power_law_1.01,0.04345600008964538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,14336,2,8,32,1,power_law_1.2,0.7923200130462646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,32,1,power_law_1.01,0.044633600115776065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,32,1,power_law_1.01,0.05247359871864319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,32,1,power_law_1.01,0.4195072174072266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,32,1,power_law_1.01,0.06471679806709289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,16384,2,8,32,1,power_law_1.2,0.17559679746627807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,32,1,power_law_1.01,1.2531392097473144
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,32,1,power_law_1.01,0.07896320223808288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,32,1,power_law_1.01,0.46823678016662595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,16384,2,8,32,1,power_law_1.2,0.03459199965000152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,16384,2,8,32,1,power_law_1.2,0.04663679897785187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,16384,2,8,32,1,power_law_1.2,0.0506816029548645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,16384,2,8,32,1,power_law_1.2,0.057631999254226685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,16384,2,8,32,1,power_law_1.2,0.06993280053138733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,16384,2,8,32,1,power_law_1.2,0.07657600045204163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,16384,2,8,32,1,power_law_1.2,0.07486720085144043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,16384,2,8,32,1,power_law_1.2,0.07901440262794494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,16384,2,8,32,1,power_law_1.2,0.08006399869918823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,16384,2,8,32,1,power_law_1.2,0.08487679958343505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,32,1,power_law_1.01,0.09269760251045227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,16384,2,8,32,1,power_law_1.2,0.08921599984169007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,16384,2,8,32,1,power_law_1.2,0.09208319783210754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,16384,2,8,32,1,power_law_1.2,0.09374079704284669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,16384,2,8,32,1,power_law_1.2,0.22858879566192628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,16384,2,8,32,1,power_law_1.2,0.10354559421539307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,16384,2,8,32,1,power_law_1.2,0.10550400018692016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,32,1,power_law_1.01,0.6000319957733155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,16384,2,8,32,1,power_law_1.2,0.11455359458923339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,16384,2,8,32,1,power_law_1.2,0.2776576042175293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,16384,2,8,32,1,power_law_1.2,0.1165887951850891
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,32,1,power_law_1.01,0.11127040386199952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,32,1,power_law_1.01,1.6505664825439452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,16384,2,8,32,1,power_law_1.2,0.04478079974651337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,16384,2,8,32,1,power_law_1.2,0.052121597528457644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,16384,2,8,32,1,power_law_1.2,0.05416319966316223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,16384,2,8,32,1,power_law_1.2,0.060038399696350095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,16384,2,8,32,1,power_law_1.2,0.06111360192298889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,16384,2,8,32,1,power_law_1.2,0.0615231990814209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,16384,2,8,32,1,power_law_1.2,0.061299198865890504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,16384,2,8,32,1,power_law_1.2,0.13875839710235596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,16384,2,8,32,1,power_law_1.2,0.06227200031280518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,16384,2,8,32,1,power_law_1.2,0.37407360076904295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,16384,2,8,32,1,power_law_1.2,0.06409599781036376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,16384,2,8,32,1,power_law_1.2,0.06558079719543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,16384,2,8,32,1,power_law_1.2,0.06638079881668091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,16384,2,8,32,1,power_law_1.2,0.06926720142364502
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,32,1,power_law_1.01,0.14884480237960815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,16384,2,8,32,1,power_law_1.2,0.16368000507354735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,14336,2,8,32,1,power_law_1.2,0.3608319997787476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,32,1,power_law_1.01,0.19067519903182983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,16384,2,8,32,1,power_law_1.2,0.07095040082931518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,16384,2,8,32,1,power_law_1.2,0.081523197889328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,16384,2,8,32,1,power_law_1.2,0.08772479891777038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,16384,2,8,32,1,power_law_1.2,0.21480960845947267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,16384,2,8,32,1,power_law_1.2,0.0998207986354828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,16384,2,8,32,1,power_law_1.2,0.09790719747543335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,16384,2,8,32,1,power_law_1.2,0.11598080396652222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,16384,2,8,32,1,power_law_1.2,0.13742079734802246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,16384,2,8,32,1,power_law_1.2,0.4672383785247803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,16384,2,8,32,1,power_law_1.2,0.25852160453796386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,16384,2,8,32,1,power_law_1.2,0.1940160036087036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,32,1,power_law_1.01,0.7805952072143555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,14336,2,8,32,1,power_law_1.2,0.8108672142028809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,16384,2,8,32,1,power_law_1.2,0.3653696060180664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,16384,2,8,32,1,power_law_1.2,0.6621695995330811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,16384,2,8,32,1,power_law_1.2,0.22783360481262208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,32,1,power_law_1.01,0.272160005569458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,32,1,power_law_1.01,0.3529664039611816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,16384,2,8,32,1,power_law_1.2,0.3248447895050049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,16384,2,8,32,1,power_law_1.2,0.46372480392456056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,32,1,power_law_1.01,0.5044352054595947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,16384,2,8,32,1,power_law_1.2,0.41635842323303224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,16384,2,8,32,1,power_law_1.2,0.6548031806945801
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,14336,2,8,32,1,power_law_1.2,1.0416000366210938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,16384,2,8,32,1,power_law_1.2,0.5949376106262207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,14336,2,8,32,1,power_law_1.2,1.0597887992858888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,32,balanced,0.08016000191370647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,32,balanced,0.08056533336639404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,32,balanced,0.08067200084527333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,32,1,power_law_1.01,1.1405632019042968
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,16384,2,8,32,1,power_law_1.2,0.026643198728561402
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,16384,2,8,32,1,power_law_1.2,0.02789120078086853
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,16384,2,8,32,1,power_law_1.2,0.027456000447273254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,16384,2,8,32,1,power_law_1.2,0.02871679961681366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,16384,2,8,32,1,power_law_1.2,0.028857600688934327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,16384,2,8,32,1,power_law_1.2,0.029811200499534608
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,16384,2,8,32,1,power_law_1.2,0.03165439963340759
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,16384,2,8,32,1,power_law_1.2,0.03527039885520935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,16384,2,8,32,1,power_law_1.2,0.03330560028553009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,32,balanced,0.09073066711425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,16384,2,8,32,1,power_law_1.2,0.03477759957313538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,32,balanced,0.2034613291422526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,16384,2,8,32,1,power_law_1.2,0.037862399220466615
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,16384,2,8,32,1,power_law_1.2,0.03723520040512085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,16384,2,8,32,1,power_law_1.2,0.03818880021572113
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,16384,2,8,32,1,power_law_1.2,0.04500479996204376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,16384,2,8,32,1,power_law_1.2,0.8689151763916015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,16384,2,8,32,1,power_law_1.2,0.043270400166511534
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,16384,2,8,32,1,power_law_1.2,0.04787839949131012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,32,balanced,0.29395200808842975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,32,balanced,0.29850133260091144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,32,balanced,0.3009973367055257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,32,balanced,0.30364267031351727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,32,balanced,0.3070720036824544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,32,balanced,0.31483733654022217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,32,balanced,0.32174932956695557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,32,balanced,0.32822932799657184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,32,balanced,0.3389706611633301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,32,balanced,0.35360534985860187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,32,balanced,0.3659733136494954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,32,balanced,0.3927253484725952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,32,balanced,0.4515093167622884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,32,balanced,0.513205329577128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,16384,2,8,32,1,power_law_1.2,0.05286399722099304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,32,balanced,0.637498656908671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,32,balanced,0.7687146663665771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,32,balanced,1.0802079836527507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,32,balanced,1.341061274210612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,16384,2,8,32,1,power_law_1.2,0.0642624020576477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,32,balanced,1.9043946266174316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,32,balanced,2.478229363759359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,16384,2,8,32,1,power_law_1.2,0.07781760096549988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,32,balanced,3.7265332539876304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,32,balanced,5.171098709106445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,32,balanced,0.04826133449872335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,32,balanced,0.0539626677831014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,32,balanced,0.04994666576385498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,32,balanced,0.07628799974918365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,32,balanced,0.1742186745007833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,32,balanced,0.27711466948191327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,32,balanced,0.2829759915669759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,32,balanced,0.28438399235407513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,32,balanced,0.2897973259290059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,32,balanced,0.2936906615893046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,32,balanced,0.302837332089742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,32,balanced,0.3130613366762797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,32,balanced,0.3215786616007487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,32,balanced,0.3380320072174072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,32,balanced,0.3571093479792277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,32,balanced,0.37332268555959064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,32,balanced,0.40725334485371906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,32,balanced,0.46996267636617023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,32,balanced,0.534437338511149
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,16384,2,8,32,1,power_law_1.2,0.0933247983455658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,32,balanced,0.6631840070088705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,32,balanced,0.7899520397186279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,32,balanced,1.1086026827494304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,32,balanced,1.369749387105306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,32,balanced,1.9350612958272297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,32,balanced,2.5090452829996743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,32,balanced,3.6414079666137695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,16384,2,8,32,1,power_law_1.2,0.7718783855438233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,16384,2,8,32,1,power_law_1.2,0.11119999885559081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,32,balanced,4.822223981221517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,32,balanced,0.07097599903742473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,32,balanced,0.07366399963696797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,32,balanced,0.07720000048478444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,32,balanced,0.08992532889048259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,32,balanced,0.1209493378798167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,32,balanced,0.20042665799458823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,32,balanced,0.20365333557128906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,32,balanced,0.2076853315035502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,32,balanced,0.21170133352279663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,32,balanced,0.21782400210698447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,32,balanced,0.22571200132369995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,32,balanced,0.23162132501602173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,32,balanced,0.2422773241996765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,32,balanced,0.2563839952150981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,32,balanced,0.2754240036010742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,32,balanced,0.289466659228007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,32,balanced,0.32019199927647907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,32,balanced,0.3806026776631673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,32,balanced,0.44125866889953613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,32,balanced,0.56004265944163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,32,balanced,0.6791733105977377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,32,balanced,0.957039992014567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,32,balanced,1.2006293137868245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,32,balanced,1.7081759770711262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,32,balanced,2.2224532763163247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,16384,2,8,32,1,power_law_1.2,0.8573184013366699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,32,balanced,3.254810651143392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,32,balanced,4.286671956380208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,16384,2,8,32,1,power_law_1.2,0.149017596244812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,32,1,power_law_1.01,0.6592319965362549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,32,power_law_1.01,0.15202560424804687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,32,power_law_1.01,0.16090879440307618
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,32,balanced,0.03156266609827677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,32,balanced,0.03136533250411352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,32,balanced,0.031530665854612984
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,32,balanced,0.033002667129039764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,14336,2,8,32,1,power_law_1.2,0.4684415817260742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,32,power_law_1.01,0.11709439754486084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,32,power_law_1.01,0.12918399572372435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,32,power_law_1.01,0.12930560111999512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,32,power_law_1.01,0.12854399681091308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,32,power_law_1.01,0.16172800064086915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,32,power_law_1.01,0.18399360179901122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,32,power_law_1.01,0.1622655987739563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,32,power_law_1.01,0.26803200244903563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,32,power_law_1.01,0.21571838855743408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,32,power_law_1.01,0.27436161041259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,32,power_law_1.01,0.28890879154205323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,32,power_law_1.01,0.27966721057891847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,32,power_law_1.01,0.2910592079162598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,32,power_law_1.01,0.2860991954803467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,32,power_law_1.01,0.2886784076690674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,32,power_law_1.01,0.2947904109954834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,32,power_law_1.01,0.30140159130096433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,32,balanced,0.04133866727352142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,32,balanced,0.06227200229962667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,32,power_law_1.01,0.3018111944198608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,32,power_law_1.01,0.30565121173858645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,32,power_law_1.01,0.30279040336608887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,32,power_law_1.01,0.31675519943237307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,32,power_law_1.01,0.30926079750061036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,32,power_law_1.01,0.3348479986190796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,32,1,power_law_1.01,1.4982463836669921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,32,power_law_1.01,0.31882240772247317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,32,power_law_1.01,0.35434880256652834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,32,power_law_1.01,0.3248703956604004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,32,power_law_1.01,0.37635838985443115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,32,power_law_1.01,0.10689280033111573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,32,power_law_1.01,0.3349695920944214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,32,power_law_1.01,0.42268800735473633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,32,power_law_1.01,0.10965759754180908
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,16384,2,8,32,1,power_law_1.2,0.19338879585266114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,32,power_law_1.01,0.3507136106491089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,32,power_law_1.01,0.11793919801712036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,32,power_law_1.01,0.37131519317626954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,32,power_law_1.01,0.12075519561767578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,32,power_law_1.01,0.41736321449279784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,32,power_law_1.01,0.1366528034210205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,32,power_law_1.01,0.5008512020111084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,32,power_law_1.01,0.1837056040763855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,32,power_law_1.01,0.19397759437561035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,32,power_law_1.01,0.2020416021347046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,32,balanced,0.0662613312403361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,32,power_law_1.01,0.4929920196533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,32,power_law_1.01,0.2072511911392212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,32,power_law_1.01,0.21172480583190917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,32,power_law_1.01,0.21941120624542237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,32,power_law_1.01,0.22910079956054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,32,power_law_1.01,0.23381121158599855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,32,power_law_1.01,0.591648006439209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,32,power_law_1.01,0.2501312017440796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,32,power_law_1.01,0.270579195022583
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,32,balanced,0.06890666484832764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,32,balanced,0.09795199831326802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,32,power_law_1.01,0.2916543960571289
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,32,balanced,0.09774399797121684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,32,balanced,0.0920960009098053
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,32,balanced,0.20806399981180826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,32,power_law_1.01,0.32855041027069093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,32,balanced,0.2165600061416626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,32,balanced,0.17941866318384805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,32,power_law_1.01,0.4013376235961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,32,power_law_1.01,0.47791361808776855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,16384,2,8,32,1,power_law_1.2,1.133241558074951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,32,power_law_1.01,0.6278207778930665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,32,power_law_1.01,0.7613120079040527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,32,power_law_1.01,0.7797376155853272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,32,power_law_1.01,0.5837823867797851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,32,power_law_1.01,0.9376768112182617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,32,power_law_1.01,0.7677824020385742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,32,balanced,0.11727466185887654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,32,power_law_1.01,1.2814208030700684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,32,power_law_1.01,0.9380800247192382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,32,power_law_1.01,1.6218751907348632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,32,power_law_1.01,1.3176639556884766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,32,power_law_1.01,1.6454208374023438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,32,power_law_1.01,1.0661760330200196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,32,balanced,0.093941330909729
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,32,balanced,0.0986293355623881
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,32,balanced,0.10531733433405559
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,32,balanced,0.11168533563613892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,32,power_law_1.01,1.3842495918273925
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,32,balanced,0.1311360001564026
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,32,balanced,0.15041066209475198
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,32,balanced,0.19122666120529175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,32,balanced,0.2348853349685669
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,32,balanced,0.3307466705640157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,32,power_law_1.01,1.9737983703613282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,32,balanced,0.4145653247833252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,32,power_law_1.01,2.3137535095214843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,32,balanced,0.5894293387730917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,32,balanced,0.7704160213470459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,32,power_law_1.01,2.527552032470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,32,power_law_1.01,2.3233280181884766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,32,power_law_1.01,2.9863616943359377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,16384,2,8,32,1,power_law_1.2,0.2720383882522583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,32,power_law_1.01,3.0947200775146486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,32,power_law_1.01,3.8282241821289062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,32,power_law_1.01,4.53759994506836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,16384,2,8,32,1,power_law_1.2,1.2500991821289062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,32,power_law_1.01,4.381510543823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,32,power_law_1.01,5.095571136474609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,32,power_law_1.2,0.15107200145721436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,32,power_law_1.2,0.16177279949188234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,32,power_law_1.01,5.956403350830078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,32,power_law_1.2,0.1614848017692566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,32,power_law_1.2,0.11520639657974244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,32,power_law_1.2,0.16104320287704468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,32,power_law_1.2,0.12818559408187866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,32,power_law_1.01,5.746579360961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,32,power_law_1.2,0.12908159494400023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,32,power_law_1.01,0.05664640069007874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,32,power_law_1.2,0.12772480249404908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,32,power_law_1.2,0.10684159994125367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,32,power_law_1.01,0.05631999969482422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,32,power_law_1.2,0.1789695978164673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,32,power_law_1.2,0.11000319719314575
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,32,power_law_1.01,0.05603839755058289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,32,power_law_1.2,0.27163519859313967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,32,power_law_1.01,0.0555840015411377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,32,power_law_1.2,0.27783679962158203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,32,power_law_1.01,0.05916159749031067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,32,power_law_1.2,0.27820160388946535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,32,power_law_1.2,0.21142399311065674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,32,power_law_1.01,0.05726720094680786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,32,power_law_1.2,0.28360960483551023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,32,power_law_1.2,0.2807424068450928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,32,power_law_1.01,0.06272640228271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,32,power_law_1.2,0.29306879043579104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,32,power_law_1.01,0.06408960223197938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,32,power_law_1.2,0.30159358978271483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,32,power_law_1.01,0.0908415973186493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,32,power_law_1.2,0.11815040111541748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,32,power_law_1.2,0.3091840028762817
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,32,power_law_1.01,0.08771839737892151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,32,power_law_1.2,0.11968640089035035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,32,power_law_1.01,0.08165119886398316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,32,power_law_1.2,0.13868160247802735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,32,power_law_1.01,0.19341440200805665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,32,power_law_1.2,0.1875264048576355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,32,power_law_1.01,0.18824959993362428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,32,power_law_1.2,0.19748480319976808
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,32,power_law_1.01,0.147871994972229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,32,power_law_1.2,0.29097599983215333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,32,power_law_1.01,0.08995199799537659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,32,power_law_1.2,0.29224960803985595
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,32,power_law_1.01,0.08852480053901672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,32,power_law_1.2,0.29697279930114745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,32,power_law_1.2,0.31946239471435545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,32,power_law_1.2,0.3025856018066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,32,power_law_1.2,0.3352576017379761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,32,power_law_1.2,0.30881919860839846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,32,power_law_1.2,0.36209280490875245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,32,power_law_1.2,0.3156800031661987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,32,power_law_1.2,0.39159040451049804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,32,power_law_1.2,0.3225663900375366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,32,power_law_1.2,0.4256319999694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,32,power_law_1.2,0.19337600469589233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,32,power_law_1.2,0.3337791919708252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,32,power_law_1.2,0.5324480056762695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,32,power_law_1.2,0.1989824056625366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,32,power_law_1.2,0.3581183910369873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,32,power_law_1.01,0.09274880290031433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,32,power_law_1.2,0.2076416015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,32,power_law_1.2,0.38661119937896726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,32,power_law_1.01,0.10196479558944702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,32,power_law_1.2,0.21365759372711182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,32,power_law_1.2,0.4165311813354492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,32,power_law_1.01,0.11025279760360718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,32,power_law_1.2,0.22162559032440185
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,32,power_law_1.01,0.13760000467300415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,32,power_law_1.2,0.5115007877349853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,32,power_law_1.2,0.2358720064163208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,32,power_law_1.01,0.1614464044570923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,32,power_law_1.2,0.24533119201660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,32,power_law_1.01,0.21984000205993653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,32,power_law_1.2,0.27024641036987307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,32,power_law_1.01,0.2720767974853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,32,power_law_1.2,0.30174078941345217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,32,power_law_1.2,0.6230016231536866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,32,power_law_1.01,0.38814079761505127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,32,power_law_1.2,0.3378495931625366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,32,power_law_1.2,0.83306884765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,32,power_law_1.01,0.4997888088226318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,32,power_law_1.2,0.40974721908569334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,32,power_law_1.2,0.48913278579711916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,32,power_law_1.2,1.0071040153503419
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,32,power_law_1.01,0.7087039947509766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,32,power_law_1.2,0.6335936069488526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,32,power_law_1.2,0.6470848083496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,32,power_law_1.2,0.8094655990600585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,32,power_law_1.2,1.3936511993408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,32,power_law_1.2,0.8102016448974609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,32,power_law_1.2,0.9943039894104004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,32,power_law_1.2,1.7313600540161134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,32,power_law_1.2,1.1145664215087892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,32,power_law_1.2,1.3977279663085938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,32,power_law_1.2,1.499071979522705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,32,power_law_1.01,1.0016127586364747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,32,power_law_1.2,2.6215232849121093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,32,power_law_1.2,2.143878364562988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,32,power_law_1.2,3.359475326538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,32,power_law_1.2,2.822412872314453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,32,power_law_1.2,1.7743295669555663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,32,power_law_1.2,2.6216896057128904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,32,power_law_1.2,5.056639862060547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,32,power_law_1.2,4.072934341430664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,32,power_law_1.2,3.5119873046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,32,power_law_1.2,5.529318237304688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,32,1,balanced,0.033914667864640556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,32,1,balanced,0.0391146664818128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,32,1,balanced,0.070592001080513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,32,power_law_1.2,0.056454402208328244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,32,1,balanced,0.09115200241406758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,32,1,balanced,0.13144532839457193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,32,1,balanced,0.20692267020543417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,32,power_law_1.2,0.05630720257759094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,32,1,balanced,0.20982933044433594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,32,1,balanced,0.2105813423792521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,32,power_law_1.2,0.055622398853302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,32,1,balanced,0.21173334121704102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,32,1,balanced,0.21363733212153116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,32,1,balanced,0.2169119914372762
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,32,power_law_1.2,0.05575680136680603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,32,1,balanced,0.21844800313313803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,32,1,balanced,0.22872533400853476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,32,1,balanced,0.2341759999593099
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,32,power_law_1.2,0.05749760270118713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,32,1,balanced,0.34381333986918133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,16384,2,8,32,1,power_law_1.2,0.3519808053970337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,32,1,balanced,0.3518506685892741
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,32,power_law_1.2,0.05557760000228882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,32,1,balanced,0.3672533432642619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,32,1,balanced,0.39673598607381183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,32,power_law_1.2,0.060844802856445314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,32,power_law_1.2,7.0959617614746096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,32,1,balanced,0.42532801628112793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,32,1,balanced,0.04110399881998698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,32,1,balanced,0.045754666129748024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,32,power_law_1.2,0.06442239880561829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,32,1,balanced,0.48395200570424396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,32,1,balanced,0.04955733319123586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,32,power_law_1.2,4.796057510375976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,32,1,balanced,0.08492799599965413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,32,1,balanced,0.1246506671110789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,32,1,balanced,0.5464906692504883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,32,power_law_1.2,0.08444799780845642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,32,1,balanced,0.7155840396881104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,32,power_law_1.2,0.08654720187187195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,32,1,balanced,0.8575413227081299
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,32,power_law_1.2,0.07693439722061157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,32,1,balanced,1.1901546319325764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,32,power_law_1.2,0.1768064022064209
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,32,power_law_1.2,0.15741440057754516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,32,1,balanced,1.500597318013509
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,32,power_law_1.2,0.14659839868545532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,32,power_law_1.2,0.10198400020599366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,32,1,balanced,2.179210662841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,32,power_law_1.2,0.08837760090827942
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,32,power_law_1.2,0.09109119772911071
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,32,power_law_1.2,0.1044160008430481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,32,1,balanced,2.9680213928222656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,32,power_law_1.2,0.11667840480804444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,32,power_law_1.2,0.14354560375213624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,32,power_law_1.2,0.17023999691009523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,32,1,balanced,0.18519999583562216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,32,power_law_1.2,0.2286207914352417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,32,1,balanced,0.187391996383667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,32,1,balanced,0.18958399693171182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,32,1,balanced,0.19131199518839517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,32,power_law_1.2,7.607852935791016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,32,power_law_1.2,0.2957119941711426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,32,1,balanced,0.1937333345413208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,32,1,balanced,0.19741332530975342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,32,1,balanced,0.20221867163976034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,32,1,power_law_1.01,0.03278720080852508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,32,1,balanced,0.21277866760889688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,32,1,balanced,0.2169653375943502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,32,1,power_law_1.01,0.03671039938926697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,32,1,balanced,0.3096533417701721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,32,1,balanced,0.31757867336273193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,32,1,power_law_1.01,0.07029759883880615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,32,1,balanced,0.3354399998982747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,32,1,balanced,0.366213321685791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,32,1,power_law_1.01,0.08912640213966369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,32,1,balanced,0.396565318107605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,32,1,balanced,0.4617493152618408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,32,1,power_law_1.01,0.11581439971923828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,32,1,balanced,0.5276106595993042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,32,1,power_law_1.01,0.14366719722747803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,32,1,balanced,0.7287253538767496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,32,1,power_law_1.01,0.16909439563751222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,32,1,balanced,0.8675093650817871
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,32,power_law_1.2,0.4237055778503418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,32,1,power_law_1.01,0.1794368028640747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,32,1,balanced,1.1882346471150715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,32,1,power_law_1.01,0.18169599771499634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,32,1,power_law_1.01,0.18707200288772582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,16384,2,8,32,1,power_law_1.2,1.2568896293640137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,16384,2,8,32,1,power_law_1.2,0.5025663852691651
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,32,1,balanced,0.03264000018437704
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,32,1,balanced,0.03169066707293192
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,32,1,balanced,0.03473600000143051
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,32,1,balanced,0.0418453315893809
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,32,1,balanced,0.06018133461475372
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,32,1,balanced,0.09709333380063374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,32,1,balanced,0.10140267014503479
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,32,1,balanced,0.10450667142868042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,32,1,balanced,0.10685333609580994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,32,1,balanced,0.1090880036354065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,32,1,balanced,0.1128000020980835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,32,1,balanced,0.11818666259447734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,32,1,balanced,0.12370666861534119
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,32,1,balanced,0.1328000028928121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,32,1,balanced,0.1606613298257192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,32,1,balanced,1.5165279706319172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,32,1,balanced,0.17103999853134155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,32,1,balanced,0.18900267283121744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,32,1,power_law_1.01,0.19668480157852172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,32,1,balanced,0.19312000274658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,32,1,balanced,0.21789334217707315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,32,1,power_law_1.01,0.2068608045578003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,32,1,balanced,0.24363199869791666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,32,1,balanced,2.188997268676758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,32,1,balanced,0.2789386709531148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,32,1,power_law_1.01,0.21529600620269776
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,32,1,balanced,0.40407999356587726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,32,1,power_law_1.01,0.23701119422912598
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,32,1,balanced,0.47355198860168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,32,1,balanced,2.853391965230306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,32,1,power_law_1.01,0.3350591897964478
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,32,power_law_1.2,0.5446464061737061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,32,1,power_law_1.01,0.03770880103111267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,32,1,power_law_1.01,0.3433727979660034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,32,1,power_law_1.01,0.04340479969978332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,32,1,power_law_1.01,0.3619647979736328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,32,power_law_1.2,0.7968319892883301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,32,1,power_law_1.01,0.0688704013824463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,32,1,power_law_1.01,0.3971776008605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,32,1,power_law_1.01,0.08509439826011658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,32,1,power_law_1.01,0.4317311763763428
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,32,power_law_1.2,1.1109951972961425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,32,1,power_law_1.01,0.11006720066070556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,32,1,power_law_1.01,0.49579520225524903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,32,1,power_law_1.01,0.13797119855880738
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,32,1,balanced,0.629040002822876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,32,1,power_law_1.01,0.5660416126251221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,32,1,power_law_1.01,0.1601472020149231
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,32,1,balanced,0.6853173573811849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,32,1,power_law_1.01,0.16475520133972169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,32,1,power_law_1.01,0.7617792129516602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,32,1,power_law_1.01,0.1686079978942871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,32,1,power_law_1.01,0.17320319414138793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,32,1,power_law_1.01,0.9310720443725586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,32,1,power_law_1.01,0.17928320169448853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,32,1,power_law_1.01,0.19248000383377076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,32,1,power_law_1.01,1.2434368133544922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,32,1,power_law_1.01,0.20565760135650635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,32,1,power_law_1.01,0.22167038917541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,32,1,power_law_1.01,1.5852224349975585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,32,1,power_law_1.01,0.30021119117736816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,32,1,balanced,0.9775199890136719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,32,1,power_law_1.01,0.3116224050521851
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,32,1,balanced,1.2743626435597737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,32,1,power_law_1.01,0.33254399299621584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,32,1,power_law_1.2,0.03306879997253418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,32,1,power_law_1.01,2.237727928161621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,32,1,power_law_1.01,0.3684544086456299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,32,1,power_law_1.2,0.04092159867286682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,32,1,power_law_1.01,0.40856318473815917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,32,1,power_law_1.2,0.07100800275802613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,32,1,power_law_1.01,0.029657599329948426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,32,1,power_law_1.01,0.4888768196105957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,32,1,power_law_1.2,0.0887615978717804
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,32,1,power_law_1.01,0.030508801341056824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,32,1,power_law_1.01,0.5774975776672363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,32,1,power_law_1.2,0.11118719577789307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,32,1,power_law_1.01,0.03375999927520752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,32,1,power_law_1.2,0.13820799589157104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,32,1,power_law_1.01,0.7562047958374023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,32,1,power_law_1.01,0.041145598888397215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,32,1,power_law_1.2,0.17326079607009887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,32,1,power_law_1.01,0.052262401580810545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,32,1,power_law_1.01,0.9160191535949707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,32,1,power_law_1.2,0.1783679962158203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,32,1,power_law_1.01,0.07467520236968994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,32,1,power_law_1.2,0.1804927945137024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,32,1,power_law_1.01,0.08972799777984619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,32,1,power_law_1.2,0.18984960317611693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,32,1,power_law_1.01,0.09461119771003723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,32,1,power_law_1.2,0.20133121013641359
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,32,1,power_law_1.01,0.09683840274810791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,32,1,power_law_1.2,0.20959360599517823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,32,1,power_law_1.01,0.10180480480194092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,32,1,power_law_1.2,0.22209279537200927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,32,1,power_law_1.01,0.10733439922332763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,32,1,power_law_1.2,0.24858241081237792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,32,1,power_law_1.01,0.11534080505371094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,32,1,power_law_1.2,0.33672959804534913
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,32,1,power_law_1.01,0.12545280456542968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,32,1,power_law_1.2,0.34437758922576905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,32,1,power_law_1.01,1.2498111724853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,32,1,power_law_1.01,0.13696639537811278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,32,1,power_law_1.2,0.3649791955947876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,32,1,power_law_1.01,0.1518272042274475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,32,1,power_law_1.2,0.3997567892074585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,32,1,power_law_1.01,0.16606080532073975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,32,1,power_law_1.01,1.582521629333496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,32,1,power_law_1.2,0.43564801216125487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,32,1,power_law_1.2,0.5001920223236084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,32,1,power_law_1.01,2.253696060180664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,32,1,power_law_1.2,0.5751616001129151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,32,1,power_law_1.2,0.7874176025390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,32,1,power_law_1.01,2.9203136444091795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,32,1,power_law_1.2,0.037913599610328676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,32,1,power_law_1.2,0.932588768005371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,32,1,power_law_1.01,2.928927993774414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,32,1,power_law_1.01,0.1952831983566284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,32,1,power_law_1.2,0.04357120096683502
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,32,1,power_law_1.01,0.18871040344238282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,32,1,power_law_1.2,0.07035520076751708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,32,1,power_law_1.2,1.2428735733032226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,32,1,power_law_1.01,0.22426879405975342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,32,1,power_law_1.2,0.08450559973716736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,32,1,power_law_1.01,0.24161279201507568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,32,1,power_law_1.2,0.10693119764328003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,32,1,power_law_1.2,1.5857728004455567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,32,1,power_law_1.01,0.29092481136322024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,32,1,power_law_1.2,0.13428479433059692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,32,1,power_law_1.2,0.16064640283584594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,32,1,power_law_1.01,0.4177663803100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,32,1,power_law_1.2,0.16473599672317504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,32,1,power_law_1.01,0.5815423965454102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,32,1,power_law_1.2,2.2459455490112306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,32,1,power_law_1.2,0.1690111994743347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,32,1,power_law_1.2,0.17406079769134522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,32,1,power_law_1.2,0.18076159954071044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,32,1,power_law_1.2,0.19274879693984986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,32,1,power_law_1.2,0.20635519027709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,32,1,power_law_1.2,0.22433919906616212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,32,1,power_law_1.2,0.3036799907684326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,32,1,power_law_1.2,0.31526401042938235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,32,1,power_law_1.2,0.029267200827598573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,32,1,power_law_1.2,0.33277440071105957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,32,1,power_law_1.2,0.02913280129432678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,32,1,power_law_1.2,0.37129600048065187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,32,1,power_law_1.2,0.033024001121521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,32,1,power_law_1.2,0.41138558387756347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,32,1,power_law_1.2,0.039340800046920775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,32,1,power_law_1.2,0.506169605255127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,16384,2,8,32,1,power_law_1.2,1.4926848411560059
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,32,1,power_law_1.2,0.050291198492050174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,32,balanced,0.047050664822260536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,32,balanced,0.048165331284205117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,32,balanced,0.048122664292653404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,32,1,power_law_1.2,0.0716159999370575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,32,balanced,0.08267199993133545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,32,balanced,0.17749333381652832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,32,balanced,0.25490667422612506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,32,1,power_law_1.2,0.09066240191459655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,32,1,power_law_1.2,2.9251455307006835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,32,1,power_law_1.01,0.5775487899780274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,32,1,power_law_1.2,0.09327999949455261
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,32,1,power_law_1.2,0.09751039743423462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,32,1,power_law_1.01,0.732147216796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,32,1,power_law_1.2,0.1002303957939148
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,32,1,power_law_1.2,0.10739200115203858
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,32,1,power_law_1.01,1.0214336395263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,32,1,power_law_1.2,0.11423360109329224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,32,1,power_law_1.2,0.5821248054504394
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,32,1,power_law_1.2,0.12599680423736573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,32,balanced,0.2581813335418701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,32,1,power_law_1.01,1.3370304107666016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,32,1,power_law_1.2,0.7697152137756348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,32,balanced,0.048885335524876915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,32,balanced,0.047770669062932335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,32,balanced,0.04814933240413666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,32,balanced,0.060826669136683144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,32,balanced,0.15254400173823038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,32,balanced,0.25070399045944214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,32,balanced,0.2553439935048421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,32,balanced,0.2569493254025777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,32,balanced,0.26293333371480304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,32,balanced,0.26495999097824097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,32,balanced,0.2754613359769185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,32,balanced,0.28167466322580975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,32,balanced,0.28890132904052734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,32,balanced,0.305893341700236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,32,balanced,0.320415993531545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,32,balanced,0.3366933266321818
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,32,1,power_law_1.2,0.13682559728622437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,32,balanced,0.3651626507441203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,32,1,power_law_1.2,0.15447039604187013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,32,balanced,0.06854933500289917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,32,balanced,0.06690133114655812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,32,balanced,0.07494399944941203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,32,1,power_law_1.2,0.16622079610824586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,32,balanced,0.08099733293056488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,32,balanced,0.1071626643339793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,32,balanced,0.18237332503000894
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,32,1,power_law_1.2,0.19598079919815065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,32,balanced,0.18495466311772665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,32,balanced,0.18913600842158
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,32,1,power_law_1.2,0.1914944052696228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,32,balanced,0.1934559941291809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,32,balanced,0.19515732924143472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,32,balanced,0.20378132661183676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,32,1,power_law_1.2,0.22704000473022462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,32,balanced,0.21167999505996704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,32,balanced,0.21902932723363241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,32,balanced,0.23444799582163492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,32,balanced,0.2490773399670919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,32,balanced,0.2653759916623433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,32,balanced,0.2617119948069255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,32,balanced,0.2943093379338582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,32,balanced,0.2634773254394531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,32,balanced,0.4248053232828776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,32,balanced,0.2671626607577006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,32,balanced,0.4866666793823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,32,balanced,0.2743733326594035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,32,1,power_law_1.2,0.9254528045654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,32,balanced,0.27956799666086835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,32,balanced,0.6040480136871338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,32,balanced,0.285535991191864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,32,balanced,0.2999733289082845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,32,balanced,0.720848004023234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,32,balanced,0.3142559925715129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,32,balanced,0.32841066519419354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,32,1,power_law_1.2,1.26046724319458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,32,balanced,1.0136213302612305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,32,balanced,0.35658665498097736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,32,balanced,0.41208000977834064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,32,balanced,1.2503466606140137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,32,balanced,0.46990398565928143
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,32,1,power_law_1.2,0.24218239784240722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,32,balanced,0.5836266676584879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,32,1,power_law_1.2,1.5949567794799804
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,32,1,power_law_1.2,0.2913343906402588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,32,balanced,0.6977226734161377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,32,balanced,0.349509318669637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,32,balanced,0.9920106728871664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,32,balanced,0.40729065736134845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,32,balanced,0.5193546613057455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,32,balanced,1.2297013600667317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,32,balanced,0.6327999830245972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,32,balanced,0.8872640132904053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,32,balanced,1.7381653785705566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,32,balanced,1.106554667154948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,32,balanced,2.263754685719808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,32,balanced,1.7706185976664226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,32,balanced,1.5885012944539387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,32,1,power_law_1.2,0.4203648090362549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,32,1,power_law_1.2,2.270412826538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,32,1,power_law_1.2,0.5195263862609864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,32,balanced,3.3695414861043296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,32,balanced,2.066645304361979
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,32,1,power_law_1.2,0.5869631767272949
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,32,1,power_law_1.2,0.7375360012054444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,32,balanced,3.0229600270589194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,32,1,power_law_1.2,2.9497600555419923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,32,balanced,4.524239857991536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,32,1,power_law_1.2,1.0254655838012696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,32,power_law_1.01,0.12851840257644653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,32,balanced,2.2898826599121094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,32,power_law_1.01,0.13630080223083496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,32,balanced,3.9834187825520835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,32,1,power_law_1.2,1.3508480072021485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,32,power_law_1.01,0.13715200424194335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,32,power_law_1.01,0.10660480260848999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,32,power_law_1.01,0.10075520277023316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,32,balanced,3.323472023010254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,32,power_law_1.01,0.11650559902191163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,32,power_law_1.01,0.10069119930267334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,32,power_law_1.01,0.1183616042137146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,32,power_law_1.01,0.10810879468917847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,32,power_law_1.01,0.11818879842758179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,32,power_law_1.01,0.10767359733581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,32,power_law_1.01,0.1603775978088379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,32,power_law_1.01,0.12020479440689087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,32,balanced,4.384954770406087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,32,power_law_1.01,0.24309120178222657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,32,power_law_1.01,0.16062719821929933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,32,power_law_1.01,0.2508224010467529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,32,power_law_1.01,0.17776639461517335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,32,power_law_1.01,0.1375488042831421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,32,power_law_1.01,0.25238399505615233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,32,power_law_1.01,0.18183679580688478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,32,power_law_1.01,0.19198720455169677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,32,power_law_1.01,0.18107520341873168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,32,power_law_1.01,0.24997119903564452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,32,power_law_1.01,0.1865664005279541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,32,power_law_1.01,0.25046401023864745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,32,power_law_1.01,0.20100479125976561
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,16384,2,8,32,1,power_law_1.2,0.6545152187347412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,32,power_law_1.01,0.25686399936676024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,32,balanced,0.029322666426499683
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,32,balanced,0.029493334392706554
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,32,balanced,0.0295413335164388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,32,power_law_1.2,0.12849279642105102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,32,balanced,0.029440000653266907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,32,balanced,0.036373332142829895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,32,balanced,0.058101331194241844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,32,power_law_1.2,0.1376512050628662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,32,balanced,0.06038400034109751
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,32,balanced,0.06262933214505513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,32,balanced,0.08724799752235413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,32,power_law_1.2,0.1388159990310669
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,32,balanced,0.08760533730189006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,32,balanced,0.08205866813659668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,32,power_law_1.2,0.13836159706115722
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,32,balanced,0.1839146614074707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,32,balanced,0.18488534291585287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,32,power_law_1.2,0.19163520336151124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,16384,2,8,32,1,power_law_1.2,1.650476837158203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,32,power_law_1.2,0.24072320461273194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,32,power_law_1.2,0.10554879903793335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,32,power_law_1.01,0.20032000541687012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,32,power_law_1.2,0.25321600437164304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,32,power_law_1.01,0.2577343940734863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,32,power_law_1.2,0.11712640523910522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,32,power_law_1.01,0.2168639898300171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,32,power_law_1.2,0.2579200029373169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,32,power_law_1.01,0.2628799915313721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,32,power_law_1.2,0.11813759803771973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,32,power_law_1.01,0.2270591974258423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,32,power_law_1.2,0.26243839263916013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,32,power_law_1.01,0.24291200637817384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,32,power_law_1.01,0.2587007999420166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,32,power_law_1.01,0.05270400047302246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,32,balanced,0.15210666259129843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,32,balanced,0.11867200334866841
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,32,balanced,0.08218133449554443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,32,balanced,0.08873066306114197
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,32,balanced,0.09314666191736858
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,32,balanced,0.09956266482671101
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,32,balanced,0.1172160009543101
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,32,balanced,0.13568533460299173
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,32,balanced,0.17080533504486084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,16384,2,8,32,1,power_law_1.2,1.6634815216064454
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,32,balanced,0.20920000473658243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,32,power_law_1.01,0.27233281135559084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,32,power_law_1.2,0.11879680156707764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,32,power_law_1.2,0.10035840272903443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,32,power_law_1.01,0.2735743999481201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,32,power_law_1.01,0.2615231990814209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,32,power_law_1.2,0.1563647985458374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,32,power_law_1.2,0.10083839893341065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,32,power_law_1.01,0.2607232093811035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,32,power_law_1.01,0.3000063896179199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,32,power_law_1.2,0.24557440280914306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,32,power_law_1.2,0.10787839889526367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,32,power_law_1.01,0.2704639911651611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,32,power_law_1.01,0.3702591896057129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,32,power_law_1.2,0.10913280248641968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,32,power_law_1.01,0.2806272029876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,32,power_law_1.01,0.4304448127746582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,32,power_law_1.2,0.1208448052406311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,32,power_law_1.01,0.2866879940032959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,32,power_law_1.01,0.30371201038360596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,32,power_law_1.2,0.17238399982452393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,32,power_law_1.01,0.5673024177551269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,32,power_law_1.01,0.32050559520721433
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,32,balanced,0.2957493265469869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,32,power_law_1.2,0.18197760581970215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,32,power_law_1.2,0.2624959945678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,32,balanced,0.36799999078114826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,32,power_law_1.01,0.34420480728149416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,32,power_law_1.2,0.17760640382766724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,32,power_law_1.2,0.27035520076751707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,32,power_law_1.01,0.38446080684661865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,32,power_law_1.01,0.05256959795951843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,32,power_law_1.2,0.1801408052444458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,32,power_law_1.2,0.2781440019607544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,32,power_law_1.01,0.4745151996612549
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,32,power_law_1.01,0.051974397897720334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,32,power_law_1.2,0.18850560188293458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,32,power_law_1.2,0.2825279951095581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,32,power_law_1.2,0.2502784013748169
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,32,power_law_1.01,0.051948797702789304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,32,power_law_1.2,0.19420160055160524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,32,power_law_1.2,0.2969280004501343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,32,power_law_1.2,0.2517375946044922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,32,power_law_1.2,0.19793920516967772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,32,power_law_1.2,0.31740798950195315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,32,power_law_1.2,0.25793280601501467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,32,power_law_1.2,0.21066880226135254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,32,power_law_1.2,0.33473920822143555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,32,power_law_1.2,0.26041600704193113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,32,power_law_1.01,0.7193280220031738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,32,power_law_1.2,0.37458560466766355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,32,power_law_1.2,0.27219200134277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,32,power_law_1.01,0.28348801136016843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,32,power_law_1.2,0.4811391830444336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,32,power_law_1.2,0.27957758903503416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,32,power_law_1.01,0.2976896047592163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,32,power_law_1.2,0.28830718994140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,32,power_law_1.01,0.31064960956573484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,32,power_law_1.2,0.5662847995758057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,32,power_law_1.01,0.5381951808929444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,32,power_law_1.01,0.32909440994262695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,32,power_law_1.2,0.736627197265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,32,power_law_1.01,0.05553920269012451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,32,power_law_1.01,0.36446080207824705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,32,power_law_1.01,0.7058752059936524
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,32,power_law_1.01,0.05440000295639038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,32,power_law_1.01,0.45216641426086424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,32,power_law_1.2,0.8867839813232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,32,power_law_1.2,0.23057279586791993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,32,power_law_1.01,0.8596287727355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,32,power_law_1.01,0.057651197910308837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,32,power_law_1.01,0.5378431797027587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,32,power_law_1.2,0.24393599033355712
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,32,power_law_1.01,0.05958399772644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,32,power_law_1.2,1.24716157913208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,32,power_law_1.2,0.2722304105758667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,32,power_law_1.01,1.1514304161071778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,32,power_law_1.01,0.7000256061553956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,32,power_law_1.01,0.07838079929351807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,32,power_law_1.2,0.3071104049682617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,32,power_law_1.01,0.8516672134399415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,32,power_law_1.2,1.6046272277832032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,32,power_law_1.01,1.5028736114501953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,32,power_law_1.01,1.1683775901794433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,32,power_law_1.2,2.3501632690429686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,32,power_law_1.01,2.164582443237305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,32,power_law_1.01,1.4827712059020997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,32,power_law_1.01,0.9957823753356934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,32,power_law_1.2,0.30163838863372805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,32,power_law_1.01,1.2638400077819825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,32,power_law_1.2,0.3776895999908447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,32,power_law_1.2,2.913875198364258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,32,power_law_1.01,2.832383918762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,32,power_law_1.2,0.4440000057220459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,32,power_law_1.01,1.8046655654907227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,32,power_law_1.2,0.5875520229339599
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,32,balanced,0.5220213333765665
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,32,balanced,0.6803306738535563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,32,power_law_1.2,0.7315648078918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,32,power_law_1.2,4.3754432678222654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,32,power_law_1.01,4.262860870361328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,32,power_law_1.01,0.0803712010383606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,32,power_law_1.01,2.1150400161743166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,32,power_law_1.01,0.07256320118904114
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,32,power_law_1.01,0.1695296049118042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,32,power_law_1.01,0.16974079608917236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,32,power_law_1.01,2.6669248580932616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,32,power_law_1.01,0.13768320083618163
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,32,power_law_1.01,0.10035200119018554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,32,power_law_1.01,5.368454360961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,32,power_law_1.2,5.925107192993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,32,power_law_1.01,0.0790719985961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,32,1,balanced,0.03143466760714849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,32,power_law_1.2,1.016038417816162
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,32,power_law_1.01,0.08665599822998046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,32,power_law_1.01,0.09230719804763794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,32,power_law_1.01,3.98089599609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,32,power_law_1.2,1.358131217956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,32,power_law_1.01,0.1026304006576538
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,32,power_law_1.01,0.1223296046257019
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,32,power_law_1.01,0.143449604511261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,32,power_law_1.01,2.32922248840332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,32,power_law_1.01,0.1877951979637146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,32,1,balanced,0.035605333745479584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,32,1,balanced,0.06018666426340739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,32,power_law_1.2,0.3258368015289307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,32,power_law_1.2,0.05139840245246887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,32,power_law_1.01,5.072415924072265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,32,power_law_1.2,0.35224959850311277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,32,1,balanced,0.037903999288876854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,32,power_law_1.2,0.05132160186767578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,32,power_law_1.2,0.3855808019638062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,32,power_law_1.2,0.05097600221633911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,32,power_law_1.01,3.4832126617431642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,32,power_law_1.2,0.4700416088104248
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,32,power_law_1.2,0.05111680030822754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,32,power_law_1.2,1.8620992660522462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,32,power_law_1.2,0.5457024097442627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,32,power_law_1.01,0.23892478942871093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,32,power_law_1.01,0.33905279636383057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,32,power_law_1.2,2.46048641204834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,32,1,balanced,0.039264000952243805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,32,1,balanced,0.047877331574757896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,32,power_law_1.01,4.672326278686524
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,32,power_law_1.01,0.45121278762817385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,32,1,power_law_1.01,0.029824000597000123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,32,power_law_1.01,0.624019193649292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,32,power_law_1.2,0.04337919950485229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,32,power_law_1.2,0.0507968008518219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,32,power_law_1.2,3.8043582916259764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,32,power_law_1.01,0.9187775611877441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,32,1,balanced,0.0806879997253418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,32,1,balanced,0.1159946620464325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,32,1,balanced,0.18127999703089395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,32,1,power_law_1.01,0.03537279963493347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,32,1,balanced,0.1829973260561625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,32,1,balanced,0.18274666865666708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,32,1,power_law_1.01,0.03716480135917664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,32,1,balanced,0.07453866799672444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,32,1,balanced,0.11215466260910034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,32,1,balanced,0.16412267088890076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,32,1,balanced,0.16791999340057373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,32,1,balanced,0.16634666919708252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,32,1,balanced,0.16870933771133423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,32,1,power_law_1.01,0.03330560028553009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,32,1,balanced,0.169322669506073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,32,1,balanced,0.17347200711568198
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,32,1,balanced,0.02926933268706004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,32,1,balanced,0.17865600188573202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,32,1,balanced,0.03126399964094162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,32,1,balanced,0.18276800711949667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,32,power_law_1.2,5.007897567749024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,32,1,balanced,0.19023466110229492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,32,power_law_1.2,0.7438464164733887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,32,power_law_1.2,0.055904000997543335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,32,power_law_1.2,0.909113597869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,32,1,balanced,0.18539732694625854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,32,1,balanced,0.18563199043273926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,32,1,balanced,0.1881706714630127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,32,1,power_law_1.01,0.06099839806556702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,32,power_law_1.2,1.2088191986083985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,32,1,power_law_1.01,0.07504640221595764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,32,1,power_law_1.01,0.09684479832649232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,32,1,power_law_1.01,0.061766397953033444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,32,1,balanced,0.03331733246644338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,32,1,balanced,0.03904533386230469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,32,power_law_1.2,1.6564992904663085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,32,1,balanced,0.27003733317057294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,32,1,balanced,0.27748799324035645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,32,1,balanced,0.29338665803273517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,32,1,balanced,0.32011733452479046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,32,power_law_1.2,0.060134398937225345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,32,1,balanced,0.34677334626515705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,32,1,balanced,0.4042453368504842
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,32,power_law_1.2,0.07947520017623902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,32,1,balanced,0.1914773384730021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,32,1,balanced,0.4593120018641154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,32,1,balanced,0.1974666714668274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,32,power_law_1.2,0.0766975998878479
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,32,power_law_1.2,0.07374079823493958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,32,power_law_1.2,0.16609280109405516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,32,1,power_law_1.01,0.028140801191329955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,32,1,power_law_1.01,0.07768959999084472
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,32,power_law_1.2,0.157043194770813
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,32,1,power_law_1.01,0.029625600576400755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,32,1,power_law_1.01,0.10083839893341065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,32,power_law_1.2,0.11496959924697876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,32,1,power_law_1.01,0.03222399950027466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,32,1,power_law_1.01,0.12880640029907225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,32,power_law_1.2,0.10239360332489014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,32,power_law_1.2,2.276767921447754
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,32,1,power_law_1.01,0.03893119990825653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,32,power_law_1.2,0.07967360019683838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,32,1,balanced,0.20251200596491495
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,32,power_law_1.2,0.08430079817771911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,32,1,balanced,0.2977866729100545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,32,1,balanced,0.3057653307914734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,32,power_law_1.2,0.09370239973068237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,32,1,balanced,0.31916799147923786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,32,1,balanced,0.34441065788269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,32,1,balanced,0.052330667773882546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,32,1,balanced,0.3698506752649943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,32,1,balanced,0.08853866656621297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,32,power_law_1.2,3.0677440643310545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,32,1,balanced,0.42211198806762695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,32,1,balanced,0.4761600097020467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,32,1,balanced,0.6276426712671915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,32,1,balanced,0.748960018157959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,32,1,balanced,0.6412853399912516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,32,1,power_law_1.01,0.04809600114822388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,32,1,balanced,0.7592480182647705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,32,1,power_law_1.01,0.06600319743156433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,32,power_law_1.2,4.386265563964844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,32,1,balanced,1.031173308690389
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,32,power_law_1.2,0.10376319885253907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,32,1,power_law_1.01,0.12391680479049683
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,32,1,balanced,0.08919466535250346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,32,1,balanced,1.3140532970428467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,32,1,balanced,0.09476266304651897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,32,1,power_law_1.01,0.14275200366973878
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,32,1,balanced,0.09497599800427754
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,32,1,balanced,0.09701866904894511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,32,1,power_law_1.01,0.14577920436859132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,32,1,balanced,1.8984427452087402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,32,1,power_law_1.01,0.15121279954910277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,32,1,power_law_1.01,0.15772160291671752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,32,1,balanced,1.03166397412618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,32,1,power_law_1.01,0.16047999858856202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,32,1,balanced,1.3035893440246582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,32,1,power_law_1.01,0.16211199760437012
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,32,1,power_law_1.01,0.07980160117149353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,32,1,power_law_1.01,0.16997120380401612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,32,power_law_1.2,0.12818559408187866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,32,1,power_law_1.01,0.08538240194320679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,32,1,balanced,1.8824052810668945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,32,1,power_law_1.01,0.18279680013656616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,32,power_law_1.2,0.15043840408325196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,32,1,power_law_1.01,0.08691840171813965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,32,1,balanced,0.10082133611043294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,32,1,balanced,0.10505066315333049
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,32,power_law_1.2,0.20892159938812255
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,32,1,power_law_1.01,0.09068160057067871
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,32,1,balanced,0.10925333698590596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,32,1,power_law_1.01,0.14865280389785768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,32,1,balanced,2.4741652806599936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,32,1,power_law_1.01,0.09507200121879578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,32,1,power_law_1.01,0.14999040365219116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,32,1,power_law_1.2,0.030976000428199767
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,32,1,power_law_1.01,0.10206719636917114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,32,1,power_law_1.01,0.15981440544128417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,32,1,power_law_1.2,0.03338240087032318
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,32,1,power_law_1.01,0.10945919752120972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,32,1,power_law_1.01,0.17018879652023317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,32,1,power_law_1.2,0.061919999122619626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,32,power_law_1.2,6.074156951904297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,32,1,power_law_1.01,0.1772096037864685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,32,1,power_law_1.2,0.07758079767227173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,32,1,power_law_1.2,0.03540480136871338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,32,1,power_law_1.01,0.1952064037322998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,32,1,power_law_1.2,0.0985472023487091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,32,1,power_law_1.2,0.03973119854927063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,32,1,power_law_1.01,0.18869119882583618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,32,1,power_law_1.2,0.1239743947982788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,32,1,power_law_1.2,0.06285439729690552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,32,1,power_law_1.01,0.20594561100006104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,32,1,balanced,0.11755733688672383
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,32,1,balanced,0.1441920002301534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,32,1,power_law_1.2,0.14729599952697753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,32,power_law_1.2,0.26345601081848147
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,32,1,balanced,0.14963199694951376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,32,1,power_law_1.01,0.2910592079162598
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,32,1,balanced,0.16665599743525186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,32,1,power_law_1.2,0.15548800230026244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,32,1,balanced,0.16875199476877847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,32,1,balanced,2.477226734161377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,32,power_law_1.2,0.3832063913345337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,32,1,balanced,0.19158399105072021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,32,1,power_law_1.2,0.15910400152206422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,32,1,balanced,0.21178666750590006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,32,1,power_law_1.01,0.12153600454330445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,32,1,power_law_1.2,0.16358400583267213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,32,1,power_law_1.01,0.1343616008758545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,32,1,power_law_1.2,0.17522560358047484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,32,1,power_law_1.01,0.14550399780273438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,32,1,power_law_1.2,0.18173439502716066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,32,1,power_law_1.01,0.2622848033905029
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,32,1,power_law_1.01,0.17127039432525634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,32,1,power_law_1.2,0.19308799505233765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,32,1,power_law_1.01,0.2718271970748901
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,32,1,power_law_1.01,0.16444159746170045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,32,1,power_law_1.2,0.2121920108795166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,32,1,power_law_1.2,0.07431039810180665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,32,1,power_law_1.01,0.289465594291687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,32,1,power_law_1.01,0.1962175965309143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,32,1,power_law_1.2,0.2913599967956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,32,1,power_law_1.2,0.09415040016174317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,32,1,power_law_1.01,0.29839999675750734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,32,1,power_law_1.01,0.208953595161438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,32,1,power_law_1.2,0.300543999671936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,32,1,power_law_1.2,0.11784319877624512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,32,1,power_law_1.01,0.3140928030014038
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,32,1,power_law_1.01,0.25080959796905516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,32,1,power_law_1.2,0.3168960094451904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,32,power_law_1.2,0.4766655921936035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,32,1,power_law_1.2,0.14086400270462035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,32,1,power_law_1.01,0.34524800777435305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,32,1,power_law_1.2,0.34644479751586915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,32,1,power_law_1.2,0.1441215991973877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,32,power_law_1.2,0.7221312046051025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,32,1,power_law_1.2,0.3786623954772949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,32,1,power_law_1.2,0.14818559885025023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,32,1,power_law_1.2,0.4370368003845215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,32,1,power_law_1.2,0.15196800231933594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,32,power_law_1.2,0.971833610534668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,32,balanced,0.027402666707833607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,32,1,power_law_1.2,0.1628543972969055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,32,balanced,0.02754666656255722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,32,balanced,0.027514666318893433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,32,balanced,0.043247997760772705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,32,1,power_law_1.2,0.1699072003364563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,32,1,balanced,0.24347732464472452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,32,1,power_law_1.2,0.18180480003356933
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,32,1,balanced,0.39366400241851807
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,32,1,power_law_1.01,0.37978239059448243
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,32,1,balanced,0.40833600362141925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,32,1,power_law_1.01,0.3767231941223145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,32,1,power_law_1.01,0.4720895767211914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,32,1,power_law_1.01,0.4337152004241943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,32,1,power_law_1.01,0.502188777923584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,32,1,power_law_1.01,0.4942272186279297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,32,1,power_law_1.2,0.5041664123535157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,32,1,power_law_1.01,0.32051839828491213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,32,1,power_law_1.2,0.027846398949623107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,32,balanced,0.05807466804981232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,32,balanced,0.05806399881839752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,32,1,power_law_1.2,0.6678271770477295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,32,1,power_law_1.01,0.3555840015411377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,32,1,power_law_1.2,0.02980479896068573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,32,1,power_law_1.2,0.20010240077972413
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,32,1,power_law_1.2,0.03218559920787811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,32,1,power_law_1.2,0.8098624229431153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,32,1,power_law_1.2,0.2651776075363159
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,32,1,balanced,0.5548906723658243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,32,1,power_law_1.2,0.27518720626831056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,32,1,power_law_1.2,1.0828288078308106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,32,1,power_law_1.2,0.291430401802063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,32,1,power_law_1.2,0.32644479274749755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,32,1,power_law_1.01,0.6328320026397705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,32,1,power_law_1.2,1.3748031616210938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,32,1,power_law_1.01,0.6552256107330322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,32,1,power_law_1.2,0.36154880523681643
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,32,1,power_law_1.01,0.8812224388122558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,32,1,power_law_1.01,0.8074815750122071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,32,1,power_law_1.01,0.42746877670288086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,32,1,power_law_1.2,1.9425024032592773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,32,1,power_law_1.2,0.03727999925613403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,32,1,power_law_1.01,0.4979135990142822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,32,1,power_law_1.01,1.0827967643737793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,32,1,power_law_1.01,1.1630080223083497
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,32,1,power_law_1.2,0.045459198951721194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,32,balanced,0.029301332930723827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,32,balanced,0.031386665999889374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,32,1,power_law_1.01,0.6645503997802734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,32,1,power_law_1.2,0.06287999749183655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,32,1,power_law_1.2,2.5206207275390624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,32,1,power_law_1.2,0.07953919768333435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,32,balanced,0.039733332892258964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,32,balanced,0.0386613334218661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,32,balanced,0.060362666845321655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,32,balanced,0.06028800209363302
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,32,1,power_law_1.2,0.0856768012046814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,32,balanced,0.06195199986298879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,32,balanced,0.06428266565004985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,32,balanced,0.06842666864395142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,32,1,power_law_1.2,0.08833280205726624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,32,1,power_law_1.2,0.09035519957542419
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,32,1,balanced,0.6009866793950399
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,32,1,power_law_1.2,0.0960640013217926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,32,1,power_law_1.2,0.10298240184783936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,32,1,balanced,0.8463679949442545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,32,balanced,0.033386667569478355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,32,balanced,0.044677332043647766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,32,balanced,0.0599839985370636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,32,balanced,0.06002666552861532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,32,1,balanced,1.0965546766916912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,32,1,power_law_1.01,0.7958847999572753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,32,balanced,0.04174399872620901
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,32,balanced,0.04276266694068909
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,32,balanced,0.04894400139649709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,32,balanced,0.049813335140546165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,32,1,power_law_1.2,0.43398399353027345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,32,1,power_law_1.01,1.0878591537475586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,32,1,power_law_1.2,0.5113088130950928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,32,1,power_law_1.01,1.3722880363464356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,32,1,power_law_1.2,0.6667391777038574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,32,balanced,0.02516799916823705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,32,1,power_law_1.01,1.3726079940795899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,32,balanced,0.06342933575312297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,32,1,power_law_1.2,0.8009535789489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,32,balanced,0.06449066599210103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,32,balanced,0.06440000236034393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,32,balanced,0.06653333206971486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,32,balanced,0.06851733227570851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,32,balanced,0.0738613357146581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,32,balanced,0.04987200101216634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,32,balanced,0.07029333213965099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,32,balanced,0.05189866820971171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,32,balanced,0.05343999962011973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,32,1,power_law_1.01,1.9406015396118164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,32,balanced,0.05426133175690969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,32,balanced,0.058117335041364036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,32,balanced,0.05991999804973602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,32,balanced,0.06217066446940104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,32,balanced,0.06858133276303609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,32,balanced,0.07384000221888225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,32,balanced,0.08098133405049641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,32,balanced,0.09117333094278972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,32,balanced,0.11538666486740112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,32,balanced,0.14095999797185263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,32,1,power_law_1.01,2.5075519561767576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,32,power_law_1.01,0.027929601073265076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,32,1,power_law_1.01,1.9577280044555665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,32,power_law_1.01,0.028070399165153505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,32,balanced,0.076773335536321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,32,1,power_law_1.2,1.0882816314697266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,32,balanced,0.07249066730340321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,32,balanced,0.0806826651096344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,32,balanced,0.08481599887212117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,32,balanced,0.09371733665466309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,32,balanced,0.10519466797510783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,32,balanced,0.1267626682917277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,32,1,power_law_1.2,1.3797951698303224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,32,balanced,0.15025066335995993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,32,balanced,0.18716265757878622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,32,balanced,0.2262880007425944
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,32,1,power_law_1.2,0.11017600297927857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,32,balanced,0.02518400053183238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,32,balanced,0.024432001014550526
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,32,balanced,0.025439999997615814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,32,1,power_law_1.2,0.1224511981010437
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,32,balanced,0.025722667574882507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,32,1,power_law_1.2,1.968511962890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,32,1,power_law_1.2,0.1367616057395935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,32,1,power_law_1.2,0.1465407967567444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,32,power_law_1.01,0.02869119942188263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,32,1,power_law_1.2,0.17209600210189818
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,32,1,power_law_1.2,0.16750719547271728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,32,1,power_law_1.2,2.556671905517578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,32,1,power_law_1.2,0.1980031967163086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,32,power_law_1.01,0.030368000268936157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,32,1,power_law_1.2,0.2120896100997925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,32,balanced,0.3137706716855367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,32,balanced,0.39661868413289386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,32,1,power_law_1.2,0.2532543897628784
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,32,balanced,0.028165332973003387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,32,balanced,0.5574080149332682
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,32,balanced,0.029680001238981884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,32,balanced,0.033514666060606636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,32,1,power_law_1.2,0.36433920860290525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,32,balanced,0.7230026721954346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,32,balanced,0.08077866832415263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,32,balanced,0.08868799606959026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,32,balanced,0.09477333227793376
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,32,1,power_law_1.2,0.45000319480895995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,32,balanced,0.10444800059000652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,32,balanced,0.12758400042851767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,32,balanced,0.14936533570289612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,32,1,power_law_1.2,0.5108992099761963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,32,balanced,0.1975733240445455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,32,balanced,0.23913600047429404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,32,balanced,0.3330186605453491
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,32,1,power_law_1.2,0.6417664051055908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,32,balanced,0.42275198300679523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,32,power_law_1.01,0.03130879998207092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,32,balanced,0.6007786591847738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,32,power_law_1.01,0.03198080062866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,32,1,power_law_1.2,0.8850624084472656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,32,power_law_1.01,0.04514560103416443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,32,balanced,0.037733333806196846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,32,power_law_1.01,0.058271998167037965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,32,power_law_1.01,0.05944960117340088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,32,balanced,1.0654773712158203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,32,1,power_law_1.2,1.1854656219482422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,32,power_law_1.01,0.062009602785110474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,32,power_law_1.01,0.06279680132865906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,32,balanced,1.3939092953999836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,32,power_law_1.01,0.037484800815582274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,32,power_law_1.01,0.06431360244750976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,32,power_law_1.01,0.04132480025291443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,32,power_law_1.01,0.06551039814949036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,32,balanced,0.17476266622543335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,32,power_law_1.01,0.04291839897632599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,32,power_law_1.01,0.06795520186424256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,32,power_law_1.01,0.04270719885826111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,32,power_law_1.01,0.07163519859313965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,32,balanced,0.7837066650390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,32,power_law_1.01,0.046079999208450316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,32,power_law_1.01,0.07432960271835327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,32,power_law_1.01,0.047443199157714847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,32,power_law_1.01,0.07985919713973999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,32,balanced,1.1436959902445476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,32,power_law_1.01,0.04955520033836365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,32,power_law_1.01,0.08636800050735474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,32,power_law_1.01,0.05043839812278748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,32,power_law_1.01,0.09327999949455261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,32,balanced,1.4991787274678547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,32,power_law_1.01,0.021766400337219237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,32,power_law_1.01,0.10325759649276733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,32,power_law_1.2,0.02922239899635315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,32,power_law_1.01,0.12685439586639405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,32,power_law_1.2,0.0290367990732193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,32,power_law_1.01,0.15425280332565308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,32,balanced,0.21611199776331583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,32,power_law_1.2,0.028697600960731505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,32,power_law_1.01,0.20096640586853026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,32,power_law_1.2,0.042745599150657655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,32,power_law_1.01,0.2497407913208008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,32,power_law_1.2,0.05496960282325745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,32,balanced,0.03812800099452337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,32,balanced,0.037717332442601524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,32,power_law_1.2,0.05567359924316406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,32,balanced,0.045882667104403176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,32,balanced,0.04571199913819631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,32,power_law_1.2,0.05868800282478333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,32,balanced,0.03923733284076055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,32,balanced,0.03805333375930786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,32,power_law_1.2,0.059084802865982056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,32,balanced,0.039077334105968475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,32,balanced,0.03956799954175949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,32,power_law_1.2,0.0616703987121582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,32,balanced,0.04181333382924398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,32,balanced,0.04381333291530609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,32,power_law_1.2,0.06517120003700257
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,32,balanced,0.04994133114814758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,32,balanced,0.05411200225353241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,32,power_law_1.2,0.06599680185317994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,32,balanced,0.06615466872851054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,32,balanced,0.07732800145943959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,32,power_law_1.2,0.07139840126037597
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,32,balanced,0.1095199982325236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,32,power_law_1.2,0.07432960271835327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,32,power_law_1.2,0.07888000011444092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,32,power_law_1.01,0.35313920974731444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,32,power_law_1.01,0.04211840033531189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,32,power_law_1.01,0.05480319857597351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,32,power_law_1.01,0.463321590423584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,32,1,power_law_1.01,2.5388479232788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,32,power_law_1.01,0.05571200251579285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,32,power_law_1.01,0.05856639742851257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,32,power_law_1.01,0.6754687786102295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,32,power_law_1.01,0.022470399737358093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,32,power_law_1.01,0.05947520136833191
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,32,power_law_1.01,0.02170239984989166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,32,power_law_1.01,0.8760000228881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,32,balanced,0.30640000104904175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,32,power_law_1.01,0.02220800071954727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,32,balanced,0.38966933886210126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,32,power_law_1.01,0.023379200696945192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,32,balanced,0.5549066861470541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,32,power_law_1.01,0.025804799795150758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,32,power_law_1.2,0.08494719862937927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,32,balanced,0.7192800045013428
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,32,power_law_1.01,0.027699199318885804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,32,power_law_1.2,0.09149439930915833
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,32,power_law_1.01,0.029894399642944335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,32,balanced,1.0497653484344482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,32,power_law_1.2,0.10396800041198731
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,32,power_law_1.01,0.03557760119438171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,32,power_law_1.2,0.12709759473800658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,32,power_law_1.2,0.03061760067939758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,32,balanced,1.378719965616862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,32,power_law_1.2,0.15016959905624389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,32,power_law_1.2,0.02975359857082367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,32,power_law_1.01,0.061401599645614625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,32,power_law_1.2,0.19240319728851318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,32,power_law_1.2,0.030220800638198854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,32,power_law_1.2,0.24746880531311036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,32,balanced,0.1420960028966268
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,32,balanced,0.19713600476582846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,32,power_law_1.01,1.2554176330566407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,32,balanced,0.25969066222508747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,32,power_law_1.2,0.3625407934188843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,32,power_law_1.01,0.05204480290412903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,32,power_law_1.01,0.05315840244293213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,32,power_law_1.01,0.036051198840141296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,32,power_law_1.01,0.05614079833030701
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,32,power_law_1.01,0.03607679903507233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,32,power_law_1.01,0.058284801244735715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,32,power_law_1.2,0.03740800023078918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,32,power_law_1.01,0.06259199976921082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,32,power_law_1.01,0.041017600893974306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,32,power_law_1.01,0.06051200032234192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,32,power_law_1.2,0.041094401478767396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,32,power_law_1.01,0.06585599780082703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,32,power_law_1.01,0.040038400888442995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,32,power_law_1.2,0.04268800020217896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,32,power_law_1.01,0.0692031979560852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,32,power_law_1.01,0.03570559918880463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,32,power_law_1.2,0.0423552006483078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,32,power_law_1.01,0.035980799794197084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,32,power_law_1.2,0.046214398741722104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,32,power_law_1.01,0.03596160113811493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,32,power_law_1.01,0.03681280016899109
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,32,power_law_1.01,0.04099839925765991
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,32,power_law_1.01,0.04293760061264038
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,32,power_law_1.01,0.05084800124168396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,32,power_law_1.2,0.044940799474716187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,32,power_law_1.01,0.058911997079849246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,32,power_law_1.2,0.05801600217819214
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,32,power_law_1.01,0.0732479989528656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,32,power_law_1.01,0.07174400091171265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,32,power_law_1.2,0.05886719822883606
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,32,power_law_1.01,0.09181439876556396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,32,power_law_1.2,0.06177279949188232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,32,power_law_1.2,0.04689280092716217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,32,power_law_1.2,0.4480127811431885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,32,power_law_1.01,0.1313024044036865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,32,power_law_1.2,0.04901120066642761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,32,power_law_1.01,1.70831356048584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,32,balanced,0.03555200000603994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,32,power_law_1.2,0.05023360252380371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,32,power_law_1.2,0.6572735786437989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,32,power_law_1.2,0.05187199711799621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,32,power_law_1.2,0.0530239999294281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,32,power_law_1.2,0.0559935986995697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,32,power_law_1.01,0.06751360297203064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,32,power_law_1.2,0.05934720039367676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,32,power_law_1.01,0.07367039918899536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,32,power_law_1.2,0.061286401748657224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,32,power_law_1.01,0.08007680177688599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,32,power_law_1.01,0.07818880081176757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,32,power_law_1.2,0.0669376015663147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,32,power_law_1.01,0.09326080083847046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,32,power_law_1.01,0.08438400030136109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,32,power_law_1.2,0.06253439784049988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,32,power_law_1.01,0.11949440240859985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,32,power_law_1.01,0.09095039963722229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,32,power_law_1.2,0.06421759724617004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,32,power_law_1.01,0.16926079988479614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,32,power_law_1.01,0.14510719776153563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,32,power_law_1.2,0.06603519916534424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,32,power_law_1.01,0.1812608003616333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,32,power_law_1.2,0.06852480173110961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,32,power_law_1.01,0.23217918872833251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,32,power_law_1.01,0.3250112056732178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,32,power_law_1.2,0.8461888313293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,32,power_law_1.2,0.023494400084018707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,32,power_law_1.01,0.41447038650512696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,32,power_law_1.2,0.023296000063419343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,32,power_law_1.2,0.022873599827289582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,32,power_law_1.2,0.07374719977378845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,32,power_law_1.2,0.0236735999584198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,32,power_law_1.2,0.07982720136642456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,32,power_law_1.01,0.1029695987701416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,32,power_law_1.2,0.024806399643421174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,32,power_law_1.01,0.12842240333557128
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,32,power_law_1.2,0.026976001262664796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,32,power_law_1.01,0.15079679489135742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,32,power_law_1.2,0.07203840017318726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,32,power_law_1.2,0.02863360047340393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,32,power_law_1.01,0.18873599767684937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,32,power_law_1.2,0.030854400992393494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,32,power_law_1.01,0.23201920986175537
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,32,power_law_1.2,0.036236798763275145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,32,power_law_1.01,0.33187839984893797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,32,power_law_1.01,0.6173823833465576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,32,balanced,0.036544000109036766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,32,balanced,0.03559466699759165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,32,power_law_1.2,0.09264640212059021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,32,power_law_1.01,0.23352959156036376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,32,power_law_1.2,0.11960320472717285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,32,power_law_1.2,0.14415359497070312
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,32,power_law_1.01,0.3124160051345825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,32,balanced,0.035391998787721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,32,balanced,0.03764266769091288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,32,balanced,0.0418453315893809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,32,balanced,0.07039999961853027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,32,balanced,0.09929600358009338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,32,balanced,0.10129066308339436
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,32,power_law_1.2,0.036601600050926206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,32,balanced,0.10322133700052898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,32,balanced,0.10708799958229065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,32,balanced,0.10941867033640544
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,32,power_law_1.2,0.036499199271202085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,32,power_law_1.01,0.43529601097106935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,32,power_law_1.2,0.04026240110397339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,32,power_law_1.01,0.6145535945892334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,32,power_law_1.2,0.03814400136470795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,32,balanced,0.08100800216197968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,32,balanced,0.12350400288899739
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,32,power_law_1.2,0.036575999855995175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,32,power_law_1.01,0.8351167678833008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,32,power_law_1.2,0.037254399061203
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,32,power_law_1.2,0.03744640052318573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,32,power_law_1.2,0.07505279779434204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,32,power_law_1.2,0.19248640537261963
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,32,power_law_1.2,0.037811198830604555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,32,power_law_1.2,0.08172159790992736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,32,power_law_1.2,0.041920000314712526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,32,power_law_1.2,0.2401279926300049
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,32,power_law_1.2,0.044787201285362246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,32,balanced,0.11250133315722148
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,32,power_law_1.2,0.05397760272026062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,32,power_law_1.2,0.3370879888534546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,32,power_law_1.2,0.06059520244598389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,32,power_law_1.01,0.783622407913208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,32,balanced,0.12462400396664937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,32,balanced,0.12988266348838806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,32,balanced,0.13179199894269308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,32,balanced,0.13345066706339517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,32,balanced,0.13621866703033447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,32,power_law_1.01,1.1306559562683105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,32,power_law_1.01,1.1973440170288085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,32,power_law_1.2,0.08718720078468323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,32,power_law_1.01,1.516915225982666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,32,balanced,0.11962667107582092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,32,power_law_1.01,1.5846464157104492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,32,balanced,0.053455998500188194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,32,balanced,0.1388800044854482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,32,balanced,0.0543146679798762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,32,balanced,0.05682666599750519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,32,balanced,0.061887999375661217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,32,power_law_1.2,0.09260159730911255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,32,balanced,0.1448906660079956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,32,power_law_1.2,0.07899519801139832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,32,power_law_1.2,0.4349055767059326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,32,balanced,0.12544000148773193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,32,balanced,0.15244266390800476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,32,power_law_1.2,0.10500479936599731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,32,balanced,0.1343946655591329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,32,balanced,0.16105600198109946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,32,balanced,0.1455946664015452
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,32,balanced,0.025445332129796345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,32,balanced,0.15408533811569214
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,32,balanced,0.025424001117547352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,32,balanced,0.1660373310248057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,32,balanced,0.18784000476201376
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,32,power_law_1.2,0.09589759707450866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,32,balanced,0.07663466533025105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,32,balanced,0.07853866616884868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,32,balanced,0.08053866525491078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,32,balanced,0.08266666531562805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,32,balanced,0.08534399668375652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,32,balanced,0.08507200082143147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,32,balanced,0.09070932865142822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,32,balanced,0.096261332432429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,32,balanced,0.10378133257230122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,32,balanced,0.11467732985814412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,32,balanced,0.12468799948692322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,32,balanced,0.13505599896113077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,32,balanced,0.15641066431999207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,32,balanced,0.2017973264058431
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,32,balanced,0.16933866341908774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,32,balanced,0.179530660311381
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,32,balanced,0.02737066646416982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,32,balanced,0.19771732886632284
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,32,balanced,0.027456000447273254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,32,balanced,0.23242133855819702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,32,balanced,0.231824000676473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,32,balanced,0.27587199211120605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,32,balanced,0.27688000599543255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,32,balanced,0.3676746686299642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,32,balanced,0.3693759838740031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,32,balanced,0.45214398701985675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,32,balanced,0.4584266742070516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,32,balanced,0.6291520198186239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,32,balanced,0.6434293190638224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,32,balanced,0.8097546895345052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,32,balanced,0.8328320185343424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,32,balanced,1.1887679894765217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,32,balanced,1.2159199714660645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,32,power_law_1.2,0.1297600030899048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,32,power_law_1.2,0.15297919511795044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,32,balanced,1.5418826738993328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,32,balanced,0.02979733298222224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,32,balanced,1.5828320185343425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,32,balanced,0.03328000009059906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,32,power_law_1.2,0.20383360385894775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,32,balanced,0.03758399933576584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,32,balanced,0.037690666814645134
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,32,balanced,0.08228800197442372
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,32,balanced,0.07875200112660725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,32,balanced,2.2771786053975425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,32,balanced,0.06620266536871593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,32,power_law_1.2,0.2603775978088379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,32,power_law_1.2,0.373088002204895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,32,balanced,3.056426684061686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,32,balanced,0.24614399671554565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,32,power_law_1.01,0.06657919883728028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,32,balanced,0.3472213347752889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,32,balanced,0.433354655901591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,32,balanced,0.6141706705093384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,32,power_law_1.2,0.6288383960723877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,32,balanced,0.7892906665802002
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,32,balanced,0.054085334142049156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,32,balanced,0.04587199787298838
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,32,balanced,0.046256000796953835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,32,balanced,0.048709332942962646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,32,balanced,0.05003733436266581
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,32,balanced,0.047882666190465294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,32,power_law_1.2,0.14458880424499512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,32,power_law_1.2,0.48859519958496095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,32,power_law_1.01,0.06680960059165955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,32,power_law_1.2,0.7005504131317138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,32,power_law_1.01,0.06739839911460876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,32,power_law_1.01,0.09891200065612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,32,balanced,2.3330772717793784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,32,power_law_1.01,0.121343994140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,32,balanced,1.139898697535197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,32,power_law_1.01,0.12357759475708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,32,power_law_1.01,0.1263808012008667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,32,balanced,1.4899786313374836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,32,power_law_1.01,0.12868479490280152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,32,balanced,3.079077402750651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,32,power_law_1.01,0.04664320051670075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,32,balanced,2.1910667419433594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,32,power_law_1.01,0.04826880097389221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,32,power_law_1.01,0.04857600033283234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,32,power_law_1.01,0.07373440265655518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,32,balanced,2.8980480829874673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,32,power_law_1.01,0.09861119985580444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,32,power_law_1.01,0.05996800065040588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,32,power_law_1.01,0.10117759704589843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,32,balanced,0.05201066533724467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,32,power_law_1.01,0.06273279786109924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,32,power_law_1.01,0.10335359573364258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,32,power_law_1.01,0.06693120002746582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,32,power_law_1.01,0.06415359973907471
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,32,power_law_1.2,0.17530239820480348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,32,power_law_1.01,0.07519360184669495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,32,power_law_1.2,0.8458880424499512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,32,balanced,0.055445333321889244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,32,balanced,0.06947200000286102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,32,power_law_1.01,0.1314239978790283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,32,balanced,0.08412266770998637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,32,balanced,0.11757333079973857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,32,power_law_1.01,0.13505280017852783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,32,balanced,0.14191466569900513
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,32,balanced,0.19646400213241577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,32,power_law_1.01,0.13742719888687133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,32,balanced,0.25563732782999676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,32,power_law_1.01,0.07708799839019775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,32,balanced,0.3612906535466512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,32,power_law_1.01,0.07921280264854431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,32,balanced,0.46775468190511066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,32,power_law_1.01,0.08060160279273987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,32,power_law_1.01,0.0831488013267517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,32,power_law_1.01,0.08465920090675354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,32,power_law_1.01,0.09060479998588562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,32,power_law_1.01,0.10729600191116333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,32,power_law_1.01,0.09671040177345276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,32,power_law_1.2,0.909119987487793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,32,power_law_1.01,0.10207359790802002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,32,power_law_1.01,0.11492480039596557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,32,power_law_1.01,0.14357119798660278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,32,power_law_1.01,0.127347195148468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,32,power_law_1.01,0.1501312017440796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,32,power_law_1.01,0.13982720375061036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,32,power_law_1.01,0.15998719930648803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,32,power_law_1.01,0.1688256025314331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,32,power_law_1.01,0.1702463984489441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,32,power_law_1.01,0.22587521076202394
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,32,power_law_1.2,0.2524224042892456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,32,power_law_1.01,0.1116096019744873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,32,power_law_1.01,0.11346559524536133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,32,power_law_1.01,0.025971201062202454
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,32,power_law_1.01,0.026265600323677064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,32,power_law_1.01,0.027014398574829103
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,32,power_law_1.01,0.027065598964691163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,32,power_law_1.01,0.1801535964012146
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,32,power_law_1.01,0.02855679988861084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,32,power_law_1.01,0.2805311918258667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,32,power_law_1.01,0.03133440017700195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,32,power_law_1.01,0.035231998562812804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,32,power_law_1.01,0.38477439880371095
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,32,power_law_1.01,0.03625600039958954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,32,power_law_1.01,0.11717120409011841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,32,power_law_1.01,0.502668809890747
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,32,power_law_1.01,0.07384960055351257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,32,power_law_1.01,0.12423039674758911
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,32,power_law_1.01,0.07237759828567505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,32,power_law_1.01,0.6893887996673584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,32,power_law_1.01,0.1310464024543762
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,32,power_law_1.01,0.05743359923362732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,32,power_law_1.01,0.14421119689941406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,32,power_law_1.01,0.04732159972190857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,32,power_law_1.01,0.881715202331543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,32,power_law_1.01,0.04355199933052063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,32,power_law_1.01,0.20171520709991456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,32,power_law_1.01,0.24003200531005858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,32,power_law_1.01,1.3629119873046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,32,power_law_1.01,0.29160959720611573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,32,power_law_1.01,0.41432957649230956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,32,power_law_1.01,1.6912191390991211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,32,power_law_1.01,0.5321407794952393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,32,power_law_1.01,0.15352959632873536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,32,power_law_1.01,0.7724480152130127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,32,power_law_1.01,0.16546560525894166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,32,power_law_1.01,0.04390400052070618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,32,power_law_1.01,2.6150272369384764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,32,power_law_1.01,0.045587199926376346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,32,power_law_1.2,1.2555264472961425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,32,power_law_1.01,0.9673343658447265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,32,power_law_1.01,0.04742400050163269
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,32,power_law_1.01,0.04855040013790131
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,32,power_law_1.01,0.05945600271224975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,32,power_law_1.01,1.4219264030456542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,32,power_law_1.01,0.06551679968833923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,32,power_law_1.01,3.31921272277832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,32,power_law_1.2,0.06740480065345764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,32,power_law_1.01,1.922368049621582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,32,power_law_1.2,0.0664896011352539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,32,power_law_1.01,0.18984960317611693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,32,power_law_1.01,0.24163200855255126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,32,power_law_1.01,0.30823678970336915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,32,power_law_1.01,2.7977535247802736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,32,power_law_1.01,0.42368640899658205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,32,power_law_1.01,0.08246399760246277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,32,power_law_1.01,0.5413504123687745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,32,power_law_1.2,1.2219008445739745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,32,power_law_1.01,0.09815679788589478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,32,power_law_1.01,0.7768959999084473
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,32,power_law_1.01,0.1344256043434143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,32,power_law_1.2,0.06700800061225891
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,32,power_law_1.01,0.16503039598464966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,32,power_law_1.2,0.09774720072746276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,32,power_law_1.01,1.0150464057922364
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,32,power_law_1.01,0.2287424087524414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,32,power_law_1.2,0.11664639711380005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,32,power_law_1.2,0.12390400171279907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,32,power_law_1.01,0.30150399208068845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,32,power_law_1.2,0.12722560167312622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,32,power_law_1.01,1.5053888320922852
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,32,power_law_1.01,0.44045438766479494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,32,power_law_1.01,0.5808000087738037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,32,power_law_1.01,1.9108991622924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,32,power_law_1.2,0.050329601764678954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,32,power_law_1.01,3.772774505615234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,32,power_law_1.2,0.050380802154541014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,32,power_law_1.2,0.04956159889698029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,32,power_law_1.2,0.07292159795761108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,32,power_law_1.2,0.09701759815216064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,32,power_law_1.01,2.9227519989013673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,32,power_law_1.2,0.10346239805221558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,32,power_law_1.2,0.10450559854507446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,32,power_law_1.2,0.11015039682388306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,32,power_law_1.2,0.11349120140075683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,32,power_law_1.2,0.11605759859085082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,32,power_law_1.2,0.12126719951629639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,32,power_law_1.2,0.05772799849510193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,32,power_law_1.01,3.852774429321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,32,power_law_1.2,0.12842240333557128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,32,power_law_1.2,0.05960959792137146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,32,power_law_1.2,0.1360383987426758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,32,power_law_1.2,0.0662015974521637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,32,power_law_1.2,0.14472320079803466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,32,power_law_1.2,0.12754559516906738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,32,power_law_1.2,0.1569599986076355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,32,power_law_1.2,0.13125120401382445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,32,power_law_1.2,0.17109760046005248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,32,power_law_1.2,0.13287680149078368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,32,power_law_1.2,0.19937920570373535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,32,power_law_1.2,0.1377344012260437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,32,power_law_1.2,0.25362560749053953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,32,power_law_1.2,0.143993604183197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,32,power_law_1.2,0.3300479888916016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,32,power_law_1.2,0.4533952236175537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,32,power_law_1.2,0.5808127880096435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,32,power_law_1.2,0.06415359973907471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,32,power_law_1.2,0.0732159972190857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,32,power_law_1.2,0.8277759552001953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,32,power_law_1.2,0.07594239711761475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,32,power_law_1.2,0.07898240089416504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,32,power_law_1.2,0.024928000569343568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,32,power_law_1.2,0.08064640164375306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,32,power_law_1.2,0.0253248006105423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,32,power_law_1.2,0.15062400102615356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,32,power_law_1.2,0.08459519743919372
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,32,power_law_1.2,0.026374399662017822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,32,power_law_1.2,0.15971839427947998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,32,power_law_1.2,0.08407040238380432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,32,power_law_1.2,0.026752001047134398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,32,power_law_1.2,0.09085440039634704
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,32,power_law_1.2,0.02773120105266571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,32,power_law_1.2,0.09530879855155945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,32,power_law_1.2,0.03076480031013489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,32,power_law_1.2,0.10501760244369507
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,32,power_law_1.2,0.03374080061912536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,32,power_law_1.2,0.11601920127868652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,32,power_law_1.2,0.034860798716545106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,32,power_law_1.2,0.13103359937667847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,32,power_law_1.2,0.0720255970954895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,32,power_law_1.2,0.14327679872512816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,32,power_law_1.2,0.07232000231742859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,32,power_law_1.2,1.1169471740722656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,32,power_law_1.2,0.17244160175323486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,32,power_law_1.2,0.0543936014175415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,32,power_law_1.2,0.22972800731658935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,32,power_law_1.2,0.04744960069656372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,32,power_law_1.2,0.28859519958496094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,32,power_law_1.2,0.043084800243377686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,32,power_law_1.2,0.4057600021362305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,32,power_law_1.2,0.04271360039710999
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,32,power_law_1.2,0.04702720046043396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,32,power_law_1.2,0.5056064128875732
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,32,power_law_1.2,0.04838399887084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,32,power_law_1.2,0.05004159808158874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,32,power_law_1.2,0.7441279888153076
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,32,power_law_1.2,0.061887997388839724
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,32,power_law_1.2,0.06871680021286011
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,32,power_law_1.2,0.09091839790344239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,32,power_law_1.2,0.17043839693069457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,32,power_law_1.2,0.9441344261169433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,32,power_law_1.2,0.1806720018386841
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,32,power_law_1.2,0.10941439867019653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,32,power_law_1.2,0.20227200984954835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,32,power_law_1.2,0.2472127914428711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,32,power_law_1.2,0.3103935956954956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,32,power_law_1.2,0.14368000030517578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,32,power_law_1.2,1.44203519821167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,32,power_law_1.2,0.45574398040771485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,32,power_law_1.2,0.5857024192810059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,32,power_law_1.2,1.587071990966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,32,power_law_1.2,1.3270719528198243
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,32,power_law_1.2,0.1775231957435608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,32,power_law_1.2,1.6242624282836915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,32,power_law_1.2,0.26309120655059814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,32,power_law_1.2,0.8059136390686035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,32,power_law_1.2,2.1694528579711916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,32,1,balanced,0.02532266577084859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,32,1,balanced,0.02736533433198929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,32,1,balanced,0.031386665999889374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,32,1,balanced,0.05799466868241628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,32,1,balanced,0.0832426647345225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,32,1,balanced,0.0828000009059906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,32,1,balanced,0.0825439989566803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,32,1,balanced,0.08346133430798848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,32,1,balanced,0.0845973292986552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,32,1,balanced,0.08545066912968953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,32,1,balanced,0.0888320008913676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,32,1,balanced,0.1358453333377838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,32,1,balanced,0.13801599542299905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,32,1,balanced,0.14230400323867798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,32,1,balanced,0.14803199966748556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,32,1,balanced,0.1546346644560496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,32,1,balanced,0.16457066933314005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,32,1,balanced,0.1835520068804423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,32,1,balanced,0.20457067092259726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,32,1,balanced,0.24542933702468872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,32,1,balanced,0.2826826572418213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,32,1,balanced,0.36793601512908936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,32,1,balanced,0.47485868136088055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,32,1,balanced,0.6665173371632894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,32,1,balanced,0.8520426750183105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,32,power_law_1.2,1.0532671928405761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,32,1,balanced,1.2508959770202637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,32,1,balanced,1.6467466354370117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,32,1,balanced,0.029669334491093952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,32,1,balanced,0.035589332381884255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,32,1,balanced,0.039701332648595176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,32,1,balanced,0.04996266464392344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,32,1,balanced,0.06805866460005443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,32,1,balanced,0.07460266848405202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,32,1,balanced,0.07620800038178761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,32,1,balanced,0.07683200140794118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,32,1,balanced,0.07894399762153625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,32,1,balanced,0.07928533355395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,32,1,balanced,0.08275199929873149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,32,1,balanced,0.11844266454378764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,32,1,balanced,0.12154666582743327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,32,1,balanced,0.12772799531618753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,32,1,balanced,0.1344586710135142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,32,1,balanced,0.13974400361378989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,32,1,balanced,0.15121066570281982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,32,1,balanced,0.17136534055074057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,32,1,balanced,0.19359999895095825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,32,1,balanced,0.2444373369216919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,32,1,balanced,0.2864266633987427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,32,1,balanced,0.39576534430185956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,32,1,balanced,0.5013440052668253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,32,1,balanced,0.7022240161895752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,32,1,balanced,0.9054186344146729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,32,power_law_1.2,1.6068416595458985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,32,1,balanced,1.321552038192749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,32,1,balanced,1.742245356241862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,32,1,power_law_1.01,0.024409599602222443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,32,1,power_law_1.01,0.03038719892501831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,32,1,power_law_1.01,0.04593279957771301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,32,1,power_law_1.01,0.054630398750305176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,32,power_law_1.2,0.3281343936920166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,32,1,power_law_1.01,0.06234880089759827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,32,1,power_law_1.01,0.0743552029132843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,32,1,power_law_1.01,0.07728000283241272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,32,1,power_law_1.01,0.07956479787826538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,32,1,power_law_1.01,0.08242560029029847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,32,1,power_law_1.01,0.08464639782905578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,32,power_law_1.2,0.3434943914413452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,32,1,power_law_1.01,0.08918399810791015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,32,1,power_law_1.01,0.029471999406814574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,32,1,power_law_1.01,0.13337600231170654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,32,1,power_law_1.01,0.032569599151611325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,32,1,power_law_1.01,0.13675520420074463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,32,1,power_law_1.01,0.04286080002784729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,32,1,power_law_1.01,0.1420032024383545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,32,1,power_law_1.01,0.05235199928283692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,32,1,power_law_1.01,0.14762879610061647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,32,1,power_law_1.01,0.06058239936828613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,32,1,power_law_1.01,0.0699455976486206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,32,1,power_law_1.01,0.07132160067558288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,32,1,power_law_1.01,0.07212799787521362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,32,1,power_law_1.01,0.07448319792747497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,32,1,power_law_1.01,0.07770879864692688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,32,1,power_law_1.01,0.08227199912071229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,32,1,power_law_1.01,0.11543680429458618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,32,1,power_law_1.01,0.11921919584274292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,32,power_law_1.2,1.7735424041748047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,32,1,power_law_1.01,0.1260543942451477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,32,1,power_law_1.01,0.13365119695663452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,32,1,power_law_1.01,0.13928960561752318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,32,1,power_law_1.01,0.15199359655380248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,32,1,power_law_1.01,0.17630720138549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,32,1,power_law_1.01,0.20021119117736816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,32,1,power_law_1.01,0.248089599609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,32,1,power_law_1.01,0.3004479885101318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,32,1,power_law_1.01,0.1541759967803955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,32,1,power_law_1.01,0.41274237632751465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,32,1,power_law_1.01,0.1661247968673706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,32,1,power_law_1.01,0.18683520555496216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,32,1,power_law_1.01,0.20881919860839843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,32,power_law_1.2,1.9768831253051757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,32,1,power_law_1.01,0.25021440982818605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,32,1,power_law_1.01,0.29180159568786623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,32,1,power_law_1.01,0.39251840114593506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,32,1,power_law_1.01,0.5088191986083984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,32,1,power_law_1.01,0.6887680053710937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,32,1,power_law_1.01,0.5198400020599365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,32,1,power_law_1.01,0.8846591949462891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,32,1,power_law_1.01,0.7236544132232666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,32,1,power_law_1.01,0.9363519668579101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,32,1,power_law_1.01,1.2819392204284668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,32,1,power_law_1.01,1.352121639251709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,32,1,power_law_1.01,1.6785663604736327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,32,1,power_law_1.2,0.02426239997148514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,32,1,power_law_1.2,0.0403328001499176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,32,1,power_law_1.01,1.7797439575195313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,32,1,power_law_1.2,0.045126399397850035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,32,1,power_law_1.2,0.029548799991607665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,32,1,power_law_1.2,0.03268480002880096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,32,1,power_law_1.2,0.035724800825119016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,32,1,power_law_1.2,0.04670720100402832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,32,1,power_law_1.2,0.056409597396850586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,32,1,power_law_1.2,0.06720640063285828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,32,1,power_law_1.2,0.07032960057258605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,32,1,power_law_1.2,0.07490559816360473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,32,1,power_law_1.2,0.07614079713821412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,32,power_law_1.2,2.6969728469848633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,32,1,power_law_1.2,0.07987840175628662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,32,1,power_law_1.2,0.08533759713172913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,32,1,power_law_1.2,0.11542400121688842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,32,1,power_law_1.2,0.11995520591735839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,32,1,power_law_1.2,0.12791039943695068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,32,1,power_law_1.2,0.13405439853668213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,32,1,power_law_1.2,0.1400320053100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,32,1,power_law_1.2,0.1530303955078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,32,1,power_law_1.2,0.17671040296554566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,32,1,power_law_1.2,0.05297920107841492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,32,1,power_law_1.2,0.20252161026000975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,32,1,power_law_1.2,0.06107519865036011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,32,1,power_law_1.2,0.07553279995918274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,32,1,power_law_1.2,0.07813760042190551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,32,1,power_law_1.2,0.08145279884338379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,32,1,power_law_1.2,0.08389760255813598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,32,1,power_law_1.2,0.08612480163574218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,32,1,power_law_1.2,0.09100160002708435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,32,1,power_law_1.2,0.13301759958267212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,32,1,power_law_1.2,0.13731199502944946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,32,1,power_law_1.2,0.14295040369033812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,32,1,power_law_1.2,0.24879999160766603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,32,1,power_law_1.2,0.1486080050468445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,32,1,power_law_1.2,0.30254080295562746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,32,1,power_law_1.2,0.41570558547973635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,32,power_law_1.2,0.47775998115539553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,32,1,power_law_1.2,0.5231167793273925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,32,1,power_law_1.2,0.726803207397461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,32,1,power_law_1.2,0.15421439409255983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,32,1,power_law_1.2,0.16670080423355102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,32,1,power_law_1.2,0.18851200342178345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,32,1,power_law_1.2,0.20959999561309814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,32,power_law_1.2,1.800364875793457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,32,balanced,0.0777706652879715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,32,balanced,0.07443200051784515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,32,1,power_law_1.2,0.25005440711975097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,32,power_law_1.2,3.001299285888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,32,1,power_law_1.2,0.29430398941040037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,32,balanced,0.07413866619269054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,32,balanced,0.14572800199190775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,32,balanced,0.18267732858657837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,32,balanced,0.2044586737950643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,32,balanced,0.20744532346725464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,32,1,power_law_1.2,0.3921152114868164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,32,balanced,0.21133333444595337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,32,balanced,0.2146773338317871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,32,balanced,0.21678932507832846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,32,balanced,0.2244960069656372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,32,balanced,0.2297333280245463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,32,balanced,0.23668799797693887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,32,balanced,0.24744532505671182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,32,balanced,0.2590026656786601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,32,1,power_law_1.2,0.5088255882263184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,32,balanced,0.2728213270505269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,32,balanced,0.2966453234354655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,32,balanced,0.35703468322753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,32,balanced,0.41844268639882404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,32,balanced,0.5751146475474039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,32,balanced,0.6962453524271647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,32,balanced,1.0179946422576904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,32,balanced,1.2791306972503662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,32,balanced,1.8256533940633137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,32,power_law_1.2,3.6257598876953123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,32,balanced,0.04598399996757507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,32,balanced,0.04790933430194855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,32,balanced,0.05013333261013031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,32,balanced,0.11454400420188904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,32,balanced,0.1567573348681132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,32,balanced,0.1851093371709188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,32,balanced,0.1893226703008016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,32,balanced,0.19132266441980997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,32,balanced,0.1954186757405599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,32,balanced,0.1996906598409017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,32,balanced,0.20788800716400146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,32,balanced,0.21609065930048624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,32,balanced,0.22398932774861655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,32,balanced,0.24048000574111938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,32,balanced,0.2564693291982015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,32,balanced,0.27136532465616864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,32,balanced,0.30191999673843384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,32,balanced,0.36287466684977215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,32,balanced,2.4390293757120767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,32,balanced,0.42798399925231934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,32,balanced,0.5860906839370728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,32,balanced,0.7142186959584554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,32,balanced,1.021173318227132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,32,balanced,3.6254612604777017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,32,balanced,1.2942826747894287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,32,balanced,1.8424800237019856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,32,balanced,4.853061358133952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,32,balanced,0.07023466626803081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,32,balanced,0.0738613357146581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,32,balanced,0.07754133145014445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,32,balanced,0.09123733639717102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,32,balanced,0.11339733004570007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,32,balanced,0.13864533106486002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,32,balanced,0.14247467120488486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,32,balanced,0.14421866337458292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,32,balanced,0.1509226659933726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,32,balanced,0.1543359955151876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,32,balanced,0.16262400150299072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,32,balanced,0.17015467087427774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,32,balanced,0.17720532417297363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,32,balanced,0.19493865966796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,32,balanced,0.20971733331680298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,32,balanced,0.2274506688117981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,32,balanced,0.25734400749206543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,32,balanced,0.31885333855946857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,32,balanced,2.4128266970316568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,32,balanced,0.3795040051142375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,32,balanced,0.532426675160726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,32,balanced,0.6514559984207153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,32,balanced,0.9189066886901855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,32,balanced,3.546346664428711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,32,1,power_law_1.2,0.68755202293396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,32,power_law_1.2,0.6355775833129883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,32,power_law_1.2,3.1500288009643556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,32,balanced,4.797045389811198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,32,power_law_1.01,0.11843199729919433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,32,power_law_1.01,0.11850240230560302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,32,power_law_1.01,0.1182528018951416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,32,balanced,1.1843039989471436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,32,power_law_1.01,0.1624959945678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,32,power_law_1.01,0.18116480112075806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,32,balanced,1.6872852643330891
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,32,power_law_1.01,0.20510079860687255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,32,power_law_1.01,0.20765440464019774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,32,balanced,2.2200800577799478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,32,power_law_1.01,0.20944640636444092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,32,power_law_1.01,0.21259520053863526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,32,power_law_1.01,0.21437439918518067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,32,power_law_1.01,0.22160000801086427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,32,1,power_law_1.2,0.9407615661621094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,32,power_law_1.01,0.2294015884399414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,32,power_law_1.01,0.23532800674438475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,32,power_law_1.01,0.24600319862365722
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,32,balanced,0.02889599899450938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,32,balanced,0.029525332152843475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,32,balanced,0.02920000006755193
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,32,balanced,0.03121600051720937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,32,balanced,0.04461866617202759
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,32,balanced,0.05516799787680308
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,32,balanced,0.05991999804973602
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,32,balanced,0.07961066563924153
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,32,balanced,0.0765173335870107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,32,balanced,3.245759963989258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,32,power_law_1.01,0.2636480093002319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,32,balanced,4.2716318766276045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,32,power_law_1.01,0.2778304100036621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,32,power_law_1.01,0.0967296004295349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,32,power_law_1.01,0.3287807941436768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,32,power_law_1.01,0.09661440253257751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,32,power_law_1.01,0.41846399307250975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,32,power_law_1.01,0.09775360226631165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,32,power_law_1.01,0.5262911796569825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,32,power_law_1.01,0.12545280456542968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,32,power_law_1.01,0.1546880006790161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,32,power_law_1.01,0.7045760154724121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,32,power_law_1.01,0.18159359693527222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,32,power_law_1.01,0.8635199546813965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,32,power_law_1.01,1.2423487663269044
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,32,balanced,0.18998932838439941
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,32,balanced,0.1650986671447754
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,32,balanced,0.15318933129310608
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,32,balanced,0.14793599645296732
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,32,balanced,0.08946133653322856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,32,power_law_1.01,1.7049728393554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,32,power_law_1.01,0.1866368055343628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,32,power_law_1.01,0.1885632038116455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,32,power_law_1.01,2.459417533874512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,32,power_law_1.01,0.19316480159759522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,32,power_law_1.01,0.19753600358963014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,32,power_law_1.01,0.20618879795074463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,32,power_law_1.01,0.21385600566864013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,32,power_law_1.01,2.993120002746582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,32,power_law_1.01,0.2225343942642212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,32,power_law_1.01,0.23939199447631837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,32,power_law_1.01,0.25788800716400145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,32,power_law_1.01,0.27674880027771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,32,power_law_1.2,1.6399744033813477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,32,power_law_1.01,0.3220288038253784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,32,power_law_1.01,0.42534399032592773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,32,power_law_1.01,4.970854568481445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,32,balanced,0.09231999516487122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,32,balanced,0.07683733105659485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,32,power_law_1.01,0.5071680068969726
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,32,balanced,0.08310399949550629
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,32,balanced,0.0879200001557668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,32,balanced,0.09389866391817729
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,32,balanced,0.11293866237004598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,32,power_law_1.01,0.6951615810394287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,32,balanced,0.1312373379866282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,32,balanced,0.17830399672190347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,32,balanced,0.2160586714744568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,32,power_law_1.01,0.9173824310302734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,32,balanced,0.3057760000228882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,32,balanced,0.39574400583902997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,32,power_law_1.01,1.2222399711608887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,32,1,power_law_1.2,0.8896512031555176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,32,power_law_1.01,0.08785920143127442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,32,power_law_1.01,6.034675216674804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,32,power_law_1.01,0.09275519847869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,32,power_law_1.01,0.10273280143737792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,32,power_law_1.01,1.6477312088012694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,32,power_law_1.01,0.10190080404281616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,32,power_law_1.01,0.1132159948348999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,32,power_law_1.01,0.1351807951927185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,32,power_law_1.01,2.3925952911376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,32,power_law_1.01,0.14076160192489623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,32,balanced,0.5516159931818644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,32,power_law_1.01,0.14248960018157958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,32,balanced,0.7523946762084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,1,32,power_law_1.2,0.11752320528030395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,32,power_law_1.01,3.257823944091797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,1,32,power_law_1.2,0.11870720386505126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,1,32,power_law_1.2,0.11743359565734864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,1,32,power_law_1.2,0.1577280044555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,1,32,power_law_1.2,0.18166400194168092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,1,32,power_law_1.2,0.2016767978668213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,32,power_law_1.01,0.1475200057029724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,1,32,power_law_1.2,0.2048896074295044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,32,power_law_1.01,0.1515455961227417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,1,32,power_law_1.2,0.20813438892364503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,32,power_law_1.01,0.1589311957359314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,1,32,power_law_1.2,0.21245439052581788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,32,power_law_1.01,0.1649791955947876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,1,32,power_law_1.2,0.21429760456085206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,32,power_law_1.01,0.17114880084991455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,1,32,power_law_1.2,0.22238080501556395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,32,power_law_1.01,0.19196799993515015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,1,32,power_law_1.2,0.2293503999710083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,32,power_law_1.01,0.21340160369873046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,32,power_law_1.01,0.04798719882965088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,1,32,power_law_1.2,0.23727359771728515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,32,power_law_1.01,0.24005119800567626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,32,power_law_1.01,0.04700160026550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,32,power_law_1.01,0.28292479515075686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,32,power_law_1.01,0.046758401393890384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,32,power_law_1.01,4.760345458984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,32,power_law_1.01,0.3547391891479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,32,power_law_1.01,0.04294399917125702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,32,power_law_1.01,0.4259007930755615
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,32,power_law_1.01,0.04152320027351379
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,32,power_law_1.01,0.05111039876937866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,32,power_law_1.01,0.6158720016479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,32,power_law_1.01,0.057606399059295654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,32,power_law_1.01,0.07374719977378845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,32,1,power_law_1.2,1.3583616256713866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,1,32,power_law_1.2,0.2507328033447266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,1,32,power_law_1.2,0.2635135889053345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,1,32,power_law_1.2,0.2787391901016235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,1,32,power_law_1.2,0.32848639488220216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,1,32,power_law_1.2,0.4479680061340332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,1,32,power_law_1.2,0.5777664184570312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,32,power_law_1.01,0.7856319904327392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,32,power_law_1.01,0.0720255970954895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,1,32,power_law_1.2,0.7624703884124756
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,32,power_law_1.01,0.16306560039520263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,1,32,power_law_1.2,0.9693568229675293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,32,power_law_1.01,0.15638400316238404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,32,power_law_1.01,0.13662079572677613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,32,power_law_1.01,0.1274623990058899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,1,32,power_law_1.2,1.3491840362548828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,32,power_law_1.01,0.08506879806518555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,32,power_law_1.01,0.08453119993209839
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,32,power_law_1.01,0.07340800166130065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,32,power_law_1.01,0.0785215973854065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,32,power_law_1.01,0.09067519903182983
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,32,power_law_1.01,0.09820160269737244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,32,power_law_1.01,0.12560640573501586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,32,power_law_1.01,0.15071359872817994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,32,power_law_1.01,0.20436480045318603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,1,32,power_law_1.2,1.9216703414916991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,32,1,power_law_1.2,1.2885184288024902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,32,power_law_1.01,1.0880255699157715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,32,power_law_1.01,6.437964630126953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,1,32,power_law_1.2,0.09751039743423462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,32,power_law_1.01,1.4204159736633302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,1,32,power_law_1.2,3.089952087402344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,32,power_law_1.01,2.0202047348022463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,32,power_law_1.01,0.2605887889862061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,1,32,power_law_1.2,3.7404159545898437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,32,power_law_1.01,2.7470783233642577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,1,32,power_law_1.2,0.09781759977340698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,1,32,power_law_1.2,0.0976639986038208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,32,power_law_1.2,4.228076934814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,1,32,power_law_1.2,0.12378239631652832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2560,8,160,1,32,power_law_1.2,0.08588160276412964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2560,8,160,1,32,power_law_1.2,0.09091839790344239
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,32,power_law_1.01,0.3810496091842651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2560,8,160,1,32,power_law_1.2,0.10108799934387207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2560,8,160,1,32,power_law_1.2,0.09672319889068604
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,32,power_law_1.01,0.49838719367980955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2560,8,160,1,32,power_law_1.2,0.11038719415664673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2560,8,160,1,32,power_law_1.2,0.13426560163497925
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,32,power_law_1.01,0.7199872016906739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2560,8,160,1,32,power_law_1.2,0.13923200368881225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2560,8,160,1,32,power_law_1.2,0.14175360202789306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,1,32,power_law_1.2,0.1526080012321472
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,32,power_law_1.01,0.958521556854248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,1,32,power_law_1.2,0.18224639892578126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,32,power_law_1.01,4.255839920043945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,1,32,power_law_1.2,0.18581759929656982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,1,32,power_law_1.2,0.18858879804611206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,1,32,power_law_1.2,0.1937343955039978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,1,32,power_law_1.2,0.19722239971160888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,1,32,power_law_1.2,0.20536320209503173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,1,32,power_law_1.2,0.215283203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,1,32,power_law_1.2,0.22406399250030518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2560,8,160,1,32,power_law_1.2,0.1463039994239807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,1,32,power_law_1.2,0.24328320026397704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,32,power_law_1.01,5.693695831298828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,1,32,power_law_1.2,0.2599551916122437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,32,1,balanced,0.03345066557327906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,32,1,balanced,0.04772266745567322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,32,1,balanced,0.06644799808661143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,1,32,power_law_1.2,0.2824959993362427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,32,1,balanced,0.09474666913350423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,32,1,balanced,0.12967999776204428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,32,1,balanced,0.14882666865984598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,1,32,power_law_1.2,0.33064320087432864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,32,1,balanced,0.14988799889882407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,32,1,balanced,0.15413332978884378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,32,1,balanced,0.1530453364054362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,1,32,power_law_1.2,0.4367743968963623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,32,1,balanced,0.1562879979610443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,32,1,balanced,0.16055466731389365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,1,32,power_law_1.2,0.559507179260254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,32,1,balanced,0.16215999921162924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,32,1,balanced,0.22827200094858804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2560,8,160,1,32,power_law_1.2,0.15128320455551147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,32,1,balanced,0.2364906668663025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,1,32,power_law_1.2,0.7778111934661865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2560,8,160,1,32,power_law_1.2,0.15815680027008056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2560,8,160,1,32,power_law_1.2,0.16851199865341188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,1,32,power_law_1.2,0.9974656105041504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2560,8,160,1,32,power_law_1.2,0.1806272029876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2560,8,160,1,32,power_law_1.2,0.20341761112213136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2560,8,160,1,32,power_law_1.2,0.22252159118652343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,1,32,power_law_1.2,1.4645055770874023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2560,8,160,1,32,power_law_1.2,0.24483840465545653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2560,8,160,1,32,power_law_1.2,0.2872512102127075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2560,8,160,1,32,power_law_1.2,0.04748800098896026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,1,32,power_law_1.2,1.7437055587768555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2560,8,160,1,32,power_law_1.2,0.04686720073223114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2560,8,160,1,32,power_law_1.2,0.3711103916168213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2560,8,160,1,32,power_law_1.2,0.04671359956264496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2560,8,160,1,32,power_law_1.2,0.44995841979980467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2560,8,160,1,32,power_law_1.2,0.039103999733924866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2560,8,160,1,32,power_law_1.2,0.6351871967315674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2560,8,160,1,32,power_law_1.2,0.04340479969978332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2560,8,160,1,32,power_law_1.2,0.05088000297546387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2560,8,160,1,32,power_law_1.2,0.8183679580688477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,1,32,power_law_1.2,6.367020797729492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2560,8,160,1,32,power_law_1.2,1.0996607780456542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,32,1,balanced,0.24156266450881958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,32,1,balanced,0.24843200047810873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,32,1,balanced,0.2617119948069255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,32,1,balanced,0.2897493243217468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,32,1,balanced,0.3163253267606099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2560,8,160,1,32,power_law_1.2,1.4978431701660155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,32,1,balanced,0.4019893407821655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,32,1,balanced,0.45677868525187176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,32,1,balanced,0.6452586650848389
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2560,8,160,1,32,power_law_1.2,0.057120001316070555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,32,1,balanced,0.7945866584777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2560,8,160,1,32,power_law_1.2,2.1685951232910154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2560,8,160,1,32,power_law_1.2,0.07279999852180481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,32,1,balanced,1.0886293252309163
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2560,8,160,1,32,power_law_1.2,0.06890239715576171
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2560,8,160,1,32,power_law_1.2,0.18050559759140014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,32,1,balanced,1.4116692543029785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2560,8,160,1,32,power_law_1.2,0.14973440170288085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2560,8,160,1,32,power_law_1.2,2.9901504516601562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2560,8,160,1,32,power_law_1.2,0.12387839555740357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,32,1,balanced,2.0286026000976562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2560,8,160,1,32,power_law_1.2,0.12305279970169067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2560,8,160,1,32,power_law_1.2,0.08300160169601441
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2560,8,160,1,32,power_law_1.2,0.08554880023002624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,32,1,balanced,2.640511989593506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,32,1,balanced,0.03749866783618927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,32,1,balanced,0.039733332892258964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2560,8,160,1,32,power_law_1.2,0.07390720248222352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,32,1,balanced,0.05874133110046387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,32,1,balanced,0.08644266923268636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,32,1,balanced,0.12427199880282085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2560,8,160,1,32,power_law_1.2,0.08011519908905029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,32,1,balanced,0.13985600074132284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,32,1,balanced,0.14020267128944397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,32,1,balanced,0.1435466706752777
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2560,8,160,1,32,power_law_1.2,0.09251840114593506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,32,1,balanced,0.14469333489735922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,32,1,balanced,0.14683733383814493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2560,8,160,1,32,power_law_1.2,0.1057919979095459
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2560,8,160,1,32,power_law_1.2,0.13878400325775148
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2560,8,160,1,32,power_law_1.2,0.15931520462036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,1,32,power_law_1.2,8.207520294189454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,32,1,power_law_1.01,0.031167998909950256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,32,1,power_law_1.01,0.03702400028705597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,32,1,power_law_1.01,0.06614400148391723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,32,1,balanced,0.15064533551534018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,32,1,power_law_1.01,0.08359040021896362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,32,1,balanced,0.1524853308995565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,32,1,balanced,0.2019253373146057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,32,1,power_law_1.01,0.10089600086212158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,32,1,balanced,0.20971733331680298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,32,1,balanced,0.21568532784779867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,32,1,power_law_1.01,0.12590080499649048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,32,1,power_law_1.01,0.13381119966506957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2560,8,160,1,32,power_law_1.2,0.22903039455413818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,32,1,power_law_1.01,0.13719680309295654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2560,8,160,1,32,power_law_1.2,0.30337278842926024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,32,1,power_law_1.01,0.1391167998313904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,32,1,power_law_1.01,0.14511359930038453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2560,8,160,1,32,power_law_1.2,0.423199987411499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,32,1,power_law_1.01,0.15672320127487183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,32,1,power_law_1.01,0.1688639998435974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,32,1,power_law_1.01,0.22359681129455566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,32,1,power_law_1.01,0.23180160522460938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,32,1,balanced,0.22435200214385986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,32,1,balanced,0.23786133527755737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,32,1,power_law_1.01,0.23914239406585694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,32,1,balanced,0.26713599761327106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,32,1,balanced,0.29651200771331787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,32,1,power_law_1.01,0.24805119037628173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,32,1,balanced,0.4073066711425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,32,1,balanced,0.4649333159128825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,32,1,power_law_1.01,0.26469759941101073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,32,1,balanced,0.6298720041910807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,32,1,power_law_1.01,0.2969919919967651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2560,8,160,1,32,power_law_1.2,4.396659088134766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,32,1,balanced,0.8004906972249349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,32,1,power_law_1.01,0.32638719081878664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,1,32,power_law_1.2,2.6701759338378905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,32,1,balanced,1.0929919878641765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,32,1,power_law_1.01,0.39634559154510496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,32,1,balanced,1.4210507074991863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,32,1,power_law_1.01,0.4901951789855957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2560,8,160,1,32,power_law_1.2,0.5738239765167237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,32,1,power_law_1.01,0.6485504150390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2560,8,160,1,32,power_law_1.2,0.7957312107086182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,1,32,power_law_1.2,3.4713409423828123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,32,1,power_law_1.01,0.7989952087402343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,32,1,power_law_1.01,1.1059200286865234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,32,1,power_law_1.01,1.4184639930725098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,32,1,balanced,2.0573813120524087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,1,32,power_law_1.2,5.526899337768555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,32,1,power_law_1.01,2.0389888763427733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,32,1,balanced,2.691258748372396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,32,1,power_law_1.01,0.03598720133304596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2560,8,160,1,32,power_law_1.2,6.174924850463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,32,1,power_law_1.01,2.6650943756103516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2560,8,160,32,1,power_law_1.2,0.031052801012992858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2560,8,160,32,1,power_law_1.2,0.03653120100498199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2560,8,160,32,1,power_law_1.2,0.05807999968528747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,32,1,power_law_1.2,0.05416319966316223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2560,8,160,32,1,power_law_1.2,0.06625279784202576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,32,1,power_law_1.2,0.06528000235557556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2560,8,160,32,1,power_law_1.2,0.08186240196228027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,32,1,power_law_1.2,0.07831040024757385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2560,8,160,32,1,power_law_1.2,0.09770879745483399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,32,1,power_law_1.2,0.09203839898109437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2560,8,160,32,1,power_law_1.2,0.12814719676971437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,32,1,power_law_1.2,0.12076799869537354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2560,8,160,32,1,power_law_1.2,0.13180160522460938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2560,8,160,32,1,power_law_1.01,0.039699199795722964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,32,1,power_law_1.2,0.12872320413589478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2560,8,160,32,1,power_law_1.01,0.06548479795455933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,32,1,power_law_1.2,0.1307136058807373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2560,8,160,32,1,power_law_1.01,0.08107519745826722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,32,1,power_law_1.2,0.13572479486465455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2560,8,160,32,1,power_law_1.01,0.09601920247077941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,32,1,power_law_1.2,0.14298880100250244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2560,8,160,32,1,power_law_1.01,0.12241920232772827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,32,1,power_law_1.2,0.15081599950790406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2560,8,160,32,1,power_law_1.01,0.12705279588699342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,32,1,power_law_1.2,0.16083840131759644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2560,8,160,32,1,power_law_1.01,0.13185919523239137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,32,1,power_law_1.2,0.19942400455474854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2560,8,160,32,1,power_law_1.01,0.13316479921340943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,32,1,power_law_1.2,0.20760960578918458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2560,8,160,32,1,power_law_1.01,0.13871359825134277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2560,8,160,32,1,power_law_1.2,0.13640320301055908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,32,1,power_law_1.2,0.21679360866546632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2560,8,160,32,1,power_law_1.01,0.14897919893264772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2560,8,160,32,1,power_law_1.2,0.14054399728775024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,32,1,power_law_1.2,0.22507519721984864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2560,8,160,32,1,power_law_1.01,0.15797760486602783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2560,8,160,32,1,power_law_1.2,0.1473088026046753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2560,8,160,32,1,power_law_1.01,0.19774080514907838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,32,1,power_law_1.2,0.24312961101531982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2560,8,160,32,1,power_law_1.2,0.15969280004501343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2560,8,160,32,1,power_law_1.01,0.2072000026702881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,32,1,power_law_1.2,0.28217599391937254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2560,8,160,32,1,power_law_1.01,0.2151103973388672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,32,1,power_law_1.2,0.32531840801239015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2560,8,160,32,1,power_law_1.01,0.2248447895050049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,32,1,power_law_1.2,0.40862717628479006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2560,8,160,32,1,power_law_1.01,0.24250240325927735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2560,8,160,32,1,power_law_1.01,0.27923200130462644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,32,1,power_law_1.2,0.4943103790283203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2560,8,160,32,1,power_law_1.01,0.314572811126709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,32,1,power_law_1.2,0.6489151954650879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2560,8,160,32,1,power_law_1.01,0.40665597915649415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2560,8,160,32,1,power_law_1.01,0.486572790145874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2560,8,160,32,1,power_law_1.2,0.17210240364074708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,32,1,power_law_1.2,0.8107135772705079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2560,8,160,32,1,power_law_1.2,0.22439041137695312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2560,8,160,32,1,power_law_1.01,0.6438720226287842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2560,8,160,32,1,power_law_1.2,0.23164160251617433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,32,1,power_law_1.2,1.1322751998901368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2560,8,160,32,1,power_law_1.01,0.8028223991394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2560,8,160,32,1,power_law_1.01,1.1248640060424804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,32,1,power_law_1.2,1.4519488334655761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2560,8,160,32,1,power_law_1.01,1.4440383911132812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2560,8,160,1,32,power_law_1.2,1.076460838317871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2560,8,160,32,1,power_law_1.2,0.2402559995651245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,32,1,power_law_1.01,2.0899072647094727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2560,8,160,32,1,power_law_1.2,0.24776959419250488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,32,power_law_1.2,4.410335922241211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2560,8,160,32,1,power_law_1.2,0.2645567893981934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,32,balanced,0.0481279989083608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,32,balanced,0.053946668903032936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,32,balanced,0.050069332122802734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2560,8,160,32,1,power_law_1.2,0.29480319023132323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,32,balanced,0.07874666651089986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,32,balanced,0.17382933696111044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2560,8,160,32,1,power_law_1.2,0.32568318843841554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,32,1,power_law_1.01,2.7337472915649412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2560,8,160,32,1,power_law_1.2,0.40218238830566405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,32,balanced,0.06991466879844666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,32,balanced,0.07469333211580913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2560,8,160,32,1,power_law_1.2,0.49793281555175783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,32,balanced,0.07853866616884868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,32,balanced,0.07914133369922638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,32,balanced,0.07655466596285503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,32,balanced,0.09222933650016785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2560,8,160,32,1,power_law_1.2,0.6472064018249511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,32,balanced,0.20310932397842407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,32,balanced,0.2938613295555115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,32,balanced,0.4334239959716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2560,8,160,32,1,power_law_1.2,0.8026240348815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,32,balanced,0.4385226567586263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,32,balanced,0.44122131665547687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,32,balanced,0.2772639989852905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,32,balanced,0.447002649307251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,32,balanced,0.41462934017181396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,32,balanced,0.4569813410441081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,32,balanced,0.42018131415049237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,32,balanced,0.466106653213501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,32,balanced,0.42457600434621173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,32,balanced,0.47393067677815753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,32,balanced,0.428607980410258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,32,balanced,0.08168533444404602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,32,balanced,0.48983999093373615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,32,balanced,0.43882668018341064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,32,balanced,0.5091413259506226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,32,balanced,0.524832010269165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,32,balanced,0.5581440130869547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,32,balanced,0.6197973489761353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,32,balanced,0.680565357208252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,32,balanced,0.8033013343811035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,32,balanced,0.9279092947642008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,32,1,power_law_1.2,1.786604881286621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,32,balanced,1.1742773056030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,32,balanced,0.4477226734161377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,32,balanced,0.45788800716400146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,32,balanced,1.5329759915669758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,32,balanced,2.021989345550537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,32,balanced,2.6331626574198403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2560,8,160,32,1,power_law_1.2,1.1095744132995606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,32,balanced,0.4732373158137004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,32,balanced,0.4912000099817912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,32,balanced,3.8310559590657554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,32,balanced,0.09123733639717102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,32,balanced,0.11973333358764648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,32,balanced,0.20153599977493286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,32,balanced,0.2927040060361226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,32,balanced,5.289072036743164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2560,8,160,32,1,power_law_1.2,1.4287360191345215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,32,power_law_1.01,0.15172480344772338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,32,balanced,0.5106666485468546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,32,power_law_1.01,0.21683199405670167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,32,balanced,0.5419306755065918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,32,balanced,0.6053386529286703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,32,power_law_1.01,0.21628160476684571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,32,balanced,0.6698506673177084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,32,balanced,0.031504000226656594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,32,balanced,0.03162133445342382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,32,balanced,0.031685332457224526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,32,power_law_1.01,0.2173759937286377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,32,power_law_1.01,0.27930240631103515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,32,balanced,0.29443200429280597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,32,balanced,0.2991200089454651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,32,balanced,0.3035680055618286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,32,power_law_1.01,0.3267199993133545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,32,balanced,0.3128533363342285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,32,balanced,0.32046399513880414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,32,balanced,0.3296159903208415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,32,power_law_1.01,0.3951936006546021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,32,balanced,0.3445173501968384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,32,balanced,0.36161601543426514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,32,balanced,0.37943466504414874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,32,power_law_1.01,0.43199357986450193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,32,balanced,0.7963679631551107
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,32,balanced,0.033359999457995095
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,32,balanced,0.04116799930731455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,32,balanced,0.06339199841022491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,32,balanced,0.9295146465301514
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,32,balanced,0.08032533526420593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,32,balanced,0.08268799881140391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,32,balanced,0.08277866741021474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,32,balanced,1.1900906562805176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,32,balanced,0.08452266454696655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,32,balanced,0.12524267037709555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,32,balanced,0.11532266934712727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,32,balanced,0.11548266808191936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,32,balanced,1.536992073059082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,32,balanced,0.26451732714970905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,32,balanced,0.21758933862050375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,32,balanced,0.18143999576568604
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,32,balanced,0.1135093371073405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,32,balanced,0.12126933534940083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,32,balanced,0.4100159804026286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,32,balanced,0.12946133812268576
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,32,balanced,0.14845866958300272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,32,power_law_1.01,0.4152383804321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,32,power_law_1.01,0.42753281593322756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,32,power_law_1.01,0.43096318244934084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,32,power_law_1.01,0.44613118171691896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,32,balanced,2.0534027417500815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2560,8,160,32,1,power_law_1.2,2.102195167541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,32,power_law_1.01,0.4549056053161621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,32,balanced,0.46904532114664715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,32,balanced,0.1668213407198588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,32,balanced,0.5291733344395956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,32,balanced,2.6538987159729004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,32,balanced,0.6487786769866943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,32,balanced,0.7641706466674805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,32,balanced,1.006229321161906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,32,balanced,3.7661654154459634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,32,balanced,1.2986400127410889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,32,balanced,1.7737760543823242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,32,balanced,0.20785067478815714
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,32,balanced,0.25092266003290814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,32,balanced,0.33295466502507526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,32,balanced,2.311840057373047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,32,balanced,0.43699200948079425
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,32,balanced,0.5933599869410197
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,32,balanced,0.811072031656901
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,32,balanced,3.3181705474853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,32,power_law_1.01,0.11637120246887207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,32,power_law_1.01,0.4668416023254395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,32,power_law_1.01,0.17356159687042236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,32,balanced,4.987818717956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,32,power_law_1.01,0.491923189163208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,32,power_law_1.01,0.1738111972808838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,32,balanced,4.401007970174153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,32,power_law_1.01,0.5136767864227295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,32,power_law_1.01,0.17363840341567993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2560,8,160,32,1,power_law_1.2,2.0506879806518556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,32,power_law_1.01,0.5400320053100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,32,power_law_1.01,0.6208255767822266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,32,power_law_1.01,0.7245696067810059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,32,power_law_1.01,0.9014719963073731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,32,1,power_law_1.2,2.750694465637207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,32,power_law_1.01,0.10687999725341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,32,power_law_1.2,0.15210239887237548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,32,power_law_1.01,0.22235519886016847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,32,1,power_law_1.2,1.6876224517822265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,32,power_law_1.01,1.0527551651000977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,32,power_law_1.2,0.2183232069015503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,32,power_law_1.2,0.11779199838638306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,32,power_law_1.01,0.2956223964691162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,32,power_law_1.2,0.21755518913269042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,32,power_law_1.2,0.173689603805542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,32,power_law_1.01,1.412320041656494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,32,power_law_1.01,0.4108799934387207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,32,power_law_1.2,0.2184000015258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,32,power_law_1.2,0.17518080472946168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,32,power_law_1.01,0.4096320152282715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,32,power_law_1.2,0.17535359859466554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,32,power_law_1.2,0.27877120971679686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,32,power_law_1.01,1.6870271682739257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,32,power_law_1.01,0.40924801826477053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,32,power_law_1.2,0.22298879623413087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,32,power_law_1.2,0.325164794921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,32,power_law_1.01,0.4120768070220947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,32,power_law_1.2,0.2891968011856079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2560,8,160,32,1,power_law_1.2,2.6889663696289063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,32,power_law_1.2,0.3957760095596313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,32,power_law_1.01,2.358732795715332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,32,power_law_1.01,0.43024640083312987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,32,power_law_1.2,0.3968512058258057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,32,power_law_1.2,0.4274496078491211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,32,power_law_1.2,0.10909440517425537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,32,power_law_1.01,0.4449920177459717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,32,power_law_1.2,0.4060671806335449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,32,power_law_1.2,0.1641535997390747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,32,power_law_1.2,0.4155327796936035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,32,power_law_1.01,0.45075201988220215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,32,power_law_1.01,3.030080032348633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,32,power_law_1.2,0.4092671871185303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,32,power_law_1.2,0.4164480209350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,32,power_law_1.01,0.16454399824142457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,32,power_law_1.01,0.45803518295288087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,32,power_law_1.2,0.41611518859863283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,32,power_law_1.2,0.44211840629577637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,32,power_law_1.01,0.16639360189437866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,32,power_law_1.01,0.48869757652282714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,32,power_law_1.2,0.42805118560791017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,32,power_law_1.2,0.4452928066253662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,32,power_law_1.01,0.16864000558853148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,32,power_law_1.01,0.5045440196990967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,32,power_law_1.01,0.056627202033996585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,32,power_law_1.2,0.4209280014038086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,32,power_law_1.01,0.1862015962600708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,32,power_law_1.01,0.07061759829521179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,32,power_law_1.01,0.5400832176208497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,32,power_law_1.01,4.521811294555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,32,power_law_1.2,0.4461184024810791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,32,power_law_1.2,0.16707199811935425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,32,power_law_1.01,0.6207488059997559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,32,power_law_1.2,0.46323838233947756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,32,power_law_1.2,0.1687551975250244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,32,power_law_1.01,0.702297592163086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,32,power_law_1.2,0.48446078300476075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,32,power_law_1.2,0.19074560403823854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,32,power_law_1.01,0.9012351989746094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,32,power_law_1.2,0.5120512008666992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,32,power_law_1.2,0.4614528179168701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,32,power_law_1.2,0.5521088123321534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,32,power_law_1.2,0.20961918830871581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,32,power_law_1.01,1.0662591934204102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,32,power_law_1.01,0.2026815891265869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,32,power_law_1.2,0.48127360343933107
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,32,power_law_1.01,0.06995199918746949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,32,power_law_1.2,0.6273087978363037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,32,power_law_1.2,0.27596800327301024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,32,power_law_1.01,0.2744319915771484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,32,power_law_1.2,0.4774911880493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,32,power_law_1.01,1.4068032264709474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,32,power_law_1.2,0.7499904155731201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,32,power_law_1.2,0.2827647924423218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,32,power_law_1.01,0.27619199752807616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,32,power_law_1.2,0.27795839309692383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,32,power_law_1.2,0.9303359985351562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,32,power_law_1.01,0.286028790473938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,32,power_law_1.01,1.7273536682128907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,32,power_law_1.2,0.27650558948516846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,32,power_law_1.2,1.0685888290405274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,32,power_law_1.01,5.890067291259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,32,power_law_1.2,0.2873791933059692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,32,power_law_1.2,0.30840320587158204
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,32,power_law_1.01,0.06940799951553345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,32,power_law_1.2,0.3083967924118042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,32,power_law_1.2,0.5168575763702392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,32,power_law_1.2,0.32798080444335936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,32,power_law_1.2,0.556550407409668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,32,power_law_1.2,0.34987521171569824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,32,power_law_1.01,0.28587520122528076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,32,power_law_1.2,0.6349823951721192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,32,power_law_1.2,0.3661952018737793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,32,power_law_1.01,0.2979327917098999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,32,power_law_1.2,0.762828779220581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,32,power_law_1.2,1.4657983779907227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,32,power_law_1.01,0.30553600788116453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,32,power_law_1.2,0.9398336410522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,32,power_law_1.01,0.07245439887046815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,32,power_law_1.01,0.31600000858306887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,32,power_law_1.2,1.8059072494506836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,32,power_law_1.01,0.07215359807014465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,32,power_law_1.2,1.1244671821594239
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,32,power_law_1.01,0.0710591971874237
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,32,power_law_1.01,0.08023040294647217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,32,power_law_1.2,1.4689472198486329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,32,power_law_1.2,2.6379392623901365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,32,power_law_1.01,2.4679424285888674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,32,power_law_1.2,1.8546688079833984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,32,power_law_1.01,0.33778560161590576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,32,power_law_1.2,3.361433410644531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,32,power_law_1.01,3.1159999847412108
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,32,power_law_1.2,0.05694720149040222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,32,power_law_1.01,0.3523904085159302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,32,power_law_1.2,2.64849910736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2560,8,160,1,32,power_law_1.2,7.343526458740234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,32,power_law_1.01,0.364518404006958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,32,1,balanced,0.03552533437808355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,32,power_law_1.01,0.07722880244255066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,32,1,balanced,0.04201599955558777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,32,1,balanced,0.07046933472156525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,32,1,balanced,0.09077866872151692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,32,power_law_1.01,0.3987519979476929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,32,1,balanced,0.13121066490809122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,32,power_law_1.01,0.07585279941558838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,32,1,balanced,0.20706667502721152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,32,1,balanced,0.29454400142033893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,32,power_law_1.01,0.47184000015258787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,32,1,balanced,0.2969706654548645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,32,1,balanced,0.29737599690755206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,32,power_law_1.01,4.484326553344727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,32,power_law_1.2,3.338387298583984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,32,power_law_1.2,5.049792098999023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,32,power_law_1.01,0.5374656200408936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,32,power_law_1.01,0.690342378616333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,32,power_law_1.2,0.4134335994720459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,32,power_law_1.01,0.835308837890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,32,power_law_1.2,0.47857279777526857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,32,power_law_1.01,0.10536320209503174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,32,power_law_1.01,1.1175488471984862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,32,power_law_1.2,4.778412628173828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,32,1,balanced,0.3056373397509257
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,32,power_law_1.01,0.09733759760856628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,32,1,balanced,0.30670400460561115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,32,1,balanced,0.31035200754801434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,32,power_law_1.2,6.437625885009766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,32,1,balanced,0.31571199496587116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,32,1,balanced,0.03965866565704346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,32,power_law_1.2,0.0694208025932312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,32,1,balanced,0.3240320086479187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,32,1,balanced,0.04608533283074697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,32,1,balanced,0.05151999990145365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,32,1,balanced,0.3327786723772685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,32,1,balanced,0.08415466547012329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,32,1,balanced,0.12382400035858154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,32,power_law_1.2,0.07004799842834472
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,32,power_law_1.2,0.06960639953613282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,32,power_law_1.2,0.5570559978485108
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,32,power_law_1.2,0.07266560196876526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,32,power_law_1.2,6.221964645385742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,32,power_law_1.2,0.07160320281982421
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,32,power_law_1.01,0.09788159728050232
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,32,power_law_1.2,0.07034879922866821
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,32,power_law_1.01,0.22344961166381835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,32,1,balanced,0.34007465839385986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,32,1,balanced,0.18745599190394083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,32,1,balanced,0.4758506615956624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,32,power_law_1.2,0.07491199970245362
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,32,power_law_1.01,0.17711360454559327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,32,1,balanced,0.5054986476898193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,32,1,balanced,0.5363519986470541
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,32,power_law_1.01,0.1500480055809021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,32,power_law_1.2,0.07247999906539918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,32,1,balanced,0.5983359813690186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,32,power_law_1.01,0.10289280414581299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,32,1,balanced,0.6542559862136841
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,32,power_law_1.2,0.07390080094337463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,32,power_law_1.01,0.1139456033706665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,32,1,balanced,0.7824479738871256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,32,power_law_1.01,1.4268223762512207
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,32,power_law_1.2,0.09203199744224548
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,32,power_law_1.01,0.12544000148773193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,32,1,balanced,1.0238933563232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,32,power_law_1.2,0.09911680221557617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,32,power_law_1.01,0.14583040475845338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,32,1,balanced,1.2779200077056885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,32,1,balanced,0.2816800077756246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,32,1,balanced,0.28428266445795697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,32,power_law_1.2,0.09745919704437256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,32,power_law_1.01,1.9769472122192382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,32,1,balanced,0.28481600681940716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,32,1,balanced,0.28684266408284503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,32,1,balanced,1.6418773333231609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,32,1,balanced,0.29047999779383343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,32,power_law_1.2,0.20734720230102538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,32,1,balanced,0.29585067431132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,32,1,balanced,0.2997066577275594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,32,1,balanced,0.30663466453552246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,32,1,balanced,2.286682605743408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,32,1,balanced,0.31435734033584595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,32,1,balanced,0.3219093283017476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,32,1,balanced,0.4300160010655721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,32,1,balanced,0.4617919921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,32,1,balanced,0.49243732293446857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,32,1,balanced,3.1484266916910806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,32,1,balanced,0.5581920146942139
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,32,power_law_1.01,0.16830079555511473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,32,1,power_law_1.01,0.033504000306129454
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,32,power_law_1.01,0.22493441104888917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,32,1,power_law_1.01,0.0365119993686676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,32,power_law_1.01,5.952767944335937
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,32,power_law_1.01,0.2724096059799194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,32,power_law_1.2,0.7208896160125733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,32,1,power_law_1.01,0.06618880033493042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,32,1,power_law_1.01,0.03829120099544525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,32,power_law_1.01,2.618988800048828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,32,1,power_law_1.01,0.089683198928833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,32,power_law_1.2,0.8700160026550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,32,1,balanced,0.029296000798543293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,32,1,balanced,0.03050133337577184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,32,1,power_law_1.01,0.12422399520874024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,32,1,balanced,0.033301333586374916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,32,1,balanced,0.03982933362325033
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,32,1,balanced,0.05825600028038025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,32,power_law_1.2,1.189516830444336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,32,1,power_law_1.01,0.16246399879455567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,32,1,balanced,0.6252373456954956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,32,1,balanced,0.7667520046234131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,32,1,power_law_1.01,0.19224319458007813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,32,power_law_1.01,3.7655361175537108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,32,1,balanced,0.9957706928253174
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,32,power_law_1.2,0.1898687958717346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,32,1,power_law_1.01,0.2259135961532593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,32,power_law_1.01,0.3815167903900146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,32,1,balanced,1.2615679899851482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,32,power_law_1.2,0.13756799697875977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,32,1,power_law_1.01,0.04257279932498932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,32,1,power_law_1.01,0.2344896078109741
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,32,power_law_1.2,0.10641280412673951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,32,1,balanced,1.6238932609558105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,32,1,power_law_1.01,0.04984320104122162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,32,1,power_law_1.01,0.23912320137023926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,32,power_law_1.2,0.11861120462417603
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,32,1,balanced,0.09710400303204854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,32,1,power_law_1.01,0.08510079979896545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,32,1,balanced,2.257045269012451
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,32,1,balanced,0.1323360006014506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,32,1,power_law_1.01,0.11697920560836791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,32,1,balanced,0.1343893309434255
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,32,1,balanced,0.13740266362826029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,32,power_law_1.2,1.4617600440979004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,32,1,power_law_1.01,0.15244799852371216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,32,1,balanced,2.9814879099527993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,32,1,power_law_1.01,0.17727999687194823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,32,power_law_1.01,0.4965504169464111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,32,1,power_law_1.01,0.20390400886535645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,32,1,power_law_1.01,0.21091198921203613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,32,1,power_law_1.01,0.24972798824310302
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,32,power_law_1.2,0.12487679719924927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,32,1,power_law_1.01,0.21660799980163575
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,32,1,balanced,0.13943466544151306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,32,power_law_1.2,0.15883519649505615
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,32,1,balanced,0.14331199725468954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,32,1,power_law_1.01,0.22641279697418212
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,32,1,balanced,0.1483573317527771
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,32,1,balanced,0.15500266353289285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,32,power_law_1.2,0.17772799730300903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,32,1,balanced,0.16302399833997092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,32,1,balanced,0.17306133111317953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,32,1,balanced,0.18382400274276733
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,32,1,balanced,0.22995734214782715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,32,power_law_1.2,2.130681610107422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,32,1,balanced,0.26953067382176715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,32,1,balanced,0.25854400793711346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,32,1,balanced,0.3142186601956685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,32,1,balanced,0.32478400071461994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,32,power_law_1.01,5.075846481323242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,32,1,power_law_1.01,0.256710410118103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,32,1,power_law_1.2,0.03367680013179779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,32,power_law_1.2,2.7454208374023437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,32,1,power_law_1.2,0.036646398901939395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,32,1,power_law_1.01,0.27185280323028566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,32,1,power_law_1.01,0.2343616008758545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,32,1,power_law_1.2,0.07037439942359924
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,32,power_law_1.2,0.22578558921813965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,32,1,power_law_1.01,0.2935424089431763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,32,1,power_law_1.01,0.2473344087600708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,32,1,power_law_1.2,0.08877440094947815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,32,power_law_1.2,0.29852800369262694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,32,1,power_law_1.01,0.31841919422149656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,32,power_law_1.01,0.7144192218780517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,32,1,power_law_1.2,0.11942399740219116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,32,1,power_law_1.01,0.3352960109710693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,32,power_law_1.2,0.42072319984436035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,32,power_law_1.2,4.074502563476562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,32,1,power_law_1.2,0.1549631953239441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,32,1,power_law_1.01,0.4525631904602051
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,32,power_law_1.01,1.0203007698059081
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,32,power_law_1.2,0.5376448154449462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,32,1,power_law_1.2,0.18266240358352662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,32,power_law_1.2,0.7523071765899658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,32,1,power_law_1.2,0.21794559955596923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,32,1,power_law_1.01,0.26175999641418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,32,1,power_law_1.2,0.2273344039916992
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,32,1,power_law_1.01,0.029382398724555968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,32,1,power_law_1.01,0.30268158912658694
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,32,1,balanced,0.4052746693293254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,32,power_law_1.2,5.298508834838867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,32,1,power_law_1.01,0.030508801341056824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,32,1,balanced,0.5511360168457031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,32,1,balanced,0.6861866315205892
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,32,1,power_law_1.01,0.03340800106525421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,32,1,power_law_1.2,0.03891839981079102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,32,1,balanced,0.9265493551890055
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,32,1,power_law_1.01,0.04074879884719849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,32,1,power_law_1.2,0.04318720102310181
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,32,1,balanced,0.9957599639892578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,32,1,power_law_1.01,0.05605120062828064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,32,1,power_law_1.01,0.08301439881324768
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,32,1,balanced,1.3474666277567546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,32,power_law_1.2,1.1151743888854981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,32,balanced,0.037615999579429626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,32,1,power_law_1.01,0.10185600519180298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,32,1,power_law_1.01,0.32444798946380615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,32,1,power_law_1.01,0.4905856132507324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,32,1,power_law_1.01,0.4169663906097412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,32,1,power_law_1.01,0.5249216079711914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,32,1,power_law_1.2,0.07024000287055969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,32,1,power_law_1.01,0.4524543762207031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,32,1,power_law_1.01,0.6028223991394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,32,1,power_law_1.2,0.08488320112228394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,32,balanced,0.0337119996547699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,32,1,power_law_1.01,0.4988416194915771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,32,balanced,0.03345600018898646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,32,1,power_law_1.2,0.23347198963165283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,32,balanced,0.03201599915822347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,32,1,power_law_1.2,0.11414400339126587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,32,balanced,0.03734400123357773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,32,balanced,0.09809600313504536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,32,balanced,0.15503467122713724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,32,balanced,0.15847466389338175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,32,1,power_law_1.2,0.14608639478683472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,32,balanced,0.15960533420244852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,32,balanced,0.1622666617234548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,32,balanced,0.1646453340848287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,32,balanced,0.17038933436075845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,32,1,power_law_1.2,0.17237759828567506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,32,balanced,0.1741866668065389
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,32,1,power_law_1.01,0.11434880495071412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,32,balanced,0.17897067467371622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,32,balanced,0.18955200910568237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,32,balanced,0.199455996354421
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,32,1,power_law_1.01,0.1185920000076294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,32,1,power_law_1.2,0.028204798698425293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,32,1,power_law_1.01,0.12249599695205689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,32,1,power_law_1.01,0.6749695777893067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,32,balanced,0.03808533400297165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,32,balanced,0.035930665830771126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,32,balanced,0.03987200061480204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,32,1,power_law_1.01,0.5921023845672607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,32,balanced,0.08701866865158081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,32,1,power_law_1.2,0.24606080055236818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,32,balanced,0.15643733739852905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,32,balanced,0.15903466939926147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,32,balanced,0.16054399808247885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,32,1,power_law_1.01,0.6724544048309327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,32,1,power_law_1.2,0.25813119411468505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,32,1,power_law_1.2,0.26973440647125246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,32,1,power_law_1.01,0.8487104415893555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,32,1,power_law_1.2,0.19952640533447266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,32,balanced,0.20945600668589273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,32,balanced,0.23029333353042603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,32,balanced,0.26929599046707153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,32,1,power_law_1.2,0.20773758888244628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,32,1,power_law_1.01,1.0051136016845703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,32,balanced,0.30690133571624756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,32,balanced,0.38442667325337726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,32,1,power_law_1.2,0.2170559883117676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,32,1,power_law_1.01,0.12985600233078004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,32,balanced,0.46048001448313397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,32,1,power_law_1.01,0.8831040382385253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,32,1,power_law_1.2,0.22278399467468263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,32,1,power_law_1.01,0.13782399892807007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,32,balanced,0.16457600394884744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,32,balanced,0.1646346648534139
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,32,1,power_law_1.01,0.14585599899291993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,32,1,power_law_1.2,0.23962879180908203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,32,1,power_law_1.01,1.020582389831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,32,1,power_law_1.01,0.15996160507202148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,32,1,power_law_1.2,0.2483839988708496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,32,1,power_law_1.2,0.2974208116531372
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,32,1,power_law_1.01,0.1738816022872925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,32,1,power_law_1.2,0.26633601188659667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,32,1,power_law_1.2,0.029600000381469725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,32,1,power_law_1.2,0.3203711986541748
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,32,1,power_law_1.01,0.18732800483703613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,32,1,power_law_1.2,0.30969600677490233
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,32,1,power_law_1.2,0.03239679932594299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,32,1,power_law_1.2,0.3391551971435547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,32,1,power_law_1.01,0.21566081047058105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,32,1,power_law_1.2,0.3300352096557617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,32,balanced,0.624944011370341
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,32,1,power_law_1.2,0.04023680090904236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,32,1,power_law_1.2,0.45469441413879397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,32,balanced,0.7756906350453695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,32,1,power_law_1.01,0.2760576009750366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,32,1,power_law_1.2,0.4188864231109619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,32,balanced,0.17084266742070517
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,32,1,power_law_1.2,0.05330560207366943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,32,balanced,0.175546665986379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,32,1,power_law_1.2,0.4897024154663086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,32,balanced,1.1237013339996338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,32,1,power_law_1.01,0.24734079837799072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,32,1,power_law_1.2,0.45813760757446287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,32,1,power_law_1.2,0.0791104018688202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,32,1,power_law_1.2,0.5252863883972168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,32,1,power_law_1.01,0.32108800411224364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,32,balanced,1.4287039438883464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,32,1,power_law_1.2,0.5049024105072022
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,32,1,power_law_1.2,0.1016767978668213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,32,1,power_law_1.01,0.3166016101837158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,32,1,power_law_1.2,0.5999743938446045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,32,balanced,2.0982185999552407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,32,1,power_law_1.01,0.4189824104309082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,32,1,power_law_1.2,0.11432960033416747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,32,1,power_law_1.01,1.3353343963623048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,32,1,power_law_1.2,0.6919360160827637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,32,1,power_law_1.2,0.11757440567016601
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,32,1,power_law_1.01,0.6253503799438477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,32,balanced,2.755407969156901
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,32,balanced,0.18023467063903809
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,32,1,power_law_1.2,0.12229119539260865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,32,balanced,0.19195733467737833
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,32,1,power_law_1.01,0.7501503944396972
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,32,1,power_law_1.2,0.1299008011817932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,32,1,power_law_1.01,1.3533696174621581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,32,1,power_law_1.2,0.609171199798584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,32,1,power_law_1.01,0.7729023933410645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,32,1,power_law_1.2,0.13692159652709962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,32,1,power_law_1.2,0.7045567989349365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,32,1,power_law_1.2,0.14570879936218262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,32,1,power_law_1.01,1.6764032363891601
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,32,1,power_law_1.01,1.060979175567627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,32,1,power_law_1.2,0.1624575972557068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,32,1,power_law_1.2,0.887929630279541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,32,1,power_law_1.2,0.8586943626403809
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,32,1,power_law_1.2,0.17496960163116454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,32,1,power_law_1.01,1.673427200317383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,32,balanced,0.20148267348607382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,32,1,power_law_1.01,1.4170559883117675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,32,balanced,0.2096959948539734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,32,1,power_law_1.2,1.0281984329223632
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,32,1,power_law_1.2,0.18843519687652588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,32,balanced,0.23081066211064658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,32,1,power_law_1.2,1.022815990447998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,32,1,power_law_1.2,0.21673600673675536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,32,1,power_law_1.2,1.3622079849243165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,32,1,power_law_1.2,0.27873280048370364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,32,1,power_law_1.2,1.3552000045776367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,32,1,power_law_1.2,0.25099520683288573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,32,1,power_law_1.2,1.6935935974121095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,32,balanced,0.04827199876308441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,32,balanced,0.04930133124192556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,32,1,power_law_1.2,1.6874752044677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,32,balanced,0.05004266897837321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,32,balanced,0.05533333122730255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,32,balanced,0.06437866886456807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,32,balanced,0.10745599865913391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,32,balanced,0.26873066027959186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,32,balanced,0.11133333047231038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,32,balanced,0.11349333326021831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,32,balanced,0.11623467008272807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,32,1,power_law_1.2,2.35863037109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,32,balanced,0.11760000387827556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,32,balanced,0.12170132994651794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,32,1,power_law_1.01,2.3412288665771483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,32,balanced,0.1288640002409617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,32,balanced,0.027445333699385326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,32,balanced,0.027488000690937042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,32,balanced,0.02740799884001414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,32,balanced,0.02743999908367793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,32,balanced,0.029482667644818623
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,32,balanced,0.033728001018365227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,32,1,power_law_1.2,0.3217407941818237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,32,1,power_law_1.2,3.0464319229125976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,32,1,power_law_1.01,3.0147327423095702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,32,power_law_1.01,0.06807680130004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,32,power_law_1.01,0.06025599837303162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,32,balanced,0.307914674282074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,32,power_law_1.01,0.0711359977722168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,32,balanced,0.38703465461730957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,32,power_law_1.01,0.06294400095939637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,32,balanced,0.4655413230260213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,32,1,power_law_1.2,2.3615360260009766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,32,balanced,0.1339306632677714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,32,power_law_1.01,0.06272000074386597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,32,balanced,0.6320266723632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,32,balanced,0.14229333400726318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,32,balanced,0.1541973352432251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,32,power_law_1.01,0.06352639794349671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,32,balanced,0.1625599960486094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,32,balanced,0.1834933360417684
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,32,balanced,0.03566933423280716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,32,balanced,0.22572267055511475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,32,balanced,0.2629599968592326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,32,1,power_law_1.2,0.3207103967666626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,32,balanced,0.34751999378204346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,32,1,power_law_1.01,2.3452096939086915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,32,balanced,0.4266666571299235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,32,1,power_law_1.2,0.42378878593444824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,32,balanced,0.5987360080083212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,32,balanced,0.7519520123799642
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,32,1,power_law_1.2,0.5648704051971436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,32,balanced,1.0728373527526855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,32,1,power_law_1.01,3.0059328079223633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,32,1,power_law_1.2,0.7630720138549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,32,balanced,1.3949653307596843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,32,power_law_1.01,0.09282559752464295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,32,power_law_1.01,0.06042879819869995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,32,power_law_1.01,0.150655996799469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,32,power_law_1.01,0.06124160289764404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,32,1,power_law_1.2,0.7856319904327392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,32,balanced,2.029066721598307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,32,power_law_1.01,0.15608960390090942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,32,power_law_1.01,0.0667136013507843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,32,power_law_1.01,0.15842560529708863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,32,power_law_1.01,0.06624640226364135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,32,power_law_1.01,0.07174400091171265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,32,power_law_1.01,0.16031999588012696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,32,power_law_1.01,0.07684479951858521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,32,power_law_1.01,0.07173759937286377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,32,power_law_1.01,0.1625920057296753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,32,balanced,0.7858239809672037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,32,power_law_1.01,0.0986303985118866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,32,power_law_1.01,0.10316159725189208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,32,power_law_1.01,0.166975998878479
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,32,balanced,0.037647999823093414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,32,balanced,1.1233440240224202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,32,power_law_1.01,0.10448000431060792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,32,power_law_1.01,0.17288320064544677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,32,power_law_1.01,0.15101439952850343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,32,power_law_1.01,0.10772479772567749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,32,1,power_law_1.2,3.0375743865966798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,32,power_law_1.01,0.11144959926605225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,32,power_law_1.01,0.11511039733886719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,32,1,power_law_1.2,1.0759360313415527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,32,balanced,2.6718292236328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,32,power_law_1.01,0.12070399522781372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,32,power_law_1.2,0.06674559712409973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,32,power_law_1.01,0.12531839609146117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,32,power_law_1.2,0.07157120108604431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,32,1,power_law_1.2,1.454899215698242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,32,power_law_1.01,0.12840960025787354
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,32,balanced,0.03995733211437861
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,32,balanced,0.03986666599909464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,32,power_law_1.2,0.07073280215263367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,32,power_law_1.01,0.13916159868240358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,32,power_law_1.01,0.17774080038070678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,32,power_law_1.2,0.0706496000289917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,32,power_law_1.01,0.15130239725112915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,32,power_law_1.01,0.18793599605560302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,32,power_law_1.2,0.0998080015182495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,32,power_law_1.01,0.16318080425262452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,32,power_law_1.01,0.19791359901428224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,32,power_law_1.2,0.1494655966758728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,32,power_law_1.01,0.1886847972869873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,32,power_law_1.01,0.2076927900314331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,32,power_law_1.01,0.22889599800109864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,32,power_law_1.01,0.03025279939174652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,32,power_law_1.01,0.26864640712738036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,32,power_law_1.01,0.029600000381469725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,32,power_law_1.01,0.31551361083984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,32,balanced,0.051216001311937966
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,32,power_law_1.01,0.029369598627090453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,32,balanced,0.1067626674969991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,32,balanced,1.439733346303304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,32,balanced,0.08940266569455464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,32,balanced,0.0823359986146291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,32,power_law_1.01,0.029446399211883544
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,32,balanced,0.06322666505972545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,32,balanced,0.059434667229652405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,32,balanced,0.06037333110968272
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,32,balanced,0.05842133363087972
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,32,balanced,0.06237866481145223
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,32,balanced,0.0687360018491745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,32,balanced,0.08288000027338664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,32,power_law_1.01,0.23416318893432617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,32,balanced,0.10520000259081523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,32,balanced,0.12603200475374857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,32,power_law_1.01,0.2758016109466553
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,32,balanced,0.17246933778127035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,32,balanced,0.2157813310623169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,32,power_law_1.01,0.36808319091796876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,32,balanced,0.30189865827560425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,32,power_law_1.2,0.060063999891281125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,32,power_law_1.01,0.40962557792663573
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,32,balanced,0.3892373243967692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,32,power_law_1.01,0.4571712017059326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,32,power_law_1.2,0.061945599317550656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,32,power_law_1.01,0.5040512084960938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,32,power_law_1.2,0.061510401964187625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,32,power_law_1.2,0.06133120059967041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,32,power_law_1.01,0.030399999022483824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,32,balanced,2.091200033823649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,32,power_law_1.2,0.06238080263137817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,32,power_law_1.2,0.062636798620224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,32,power_law_1.01,0.031590399146080014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,32,power_law_1.2,0.06618239879608154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,32,power_law_1.2,0.1556607961654663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,32,power_law_1.2,0.09244800209999085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,32,power_law_1.01,0.03454720079898834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,32,power_law_1.2,0.06657919883728028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,32,power_law_1.2,0.15624959468841554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,32,balanced,2.738880157470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,32,power_law_1.2,0.1508288025856018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,32,power_law_1.2,0.07251840233802795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,32,power_law_1.2,0.15997439622879028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,32,power_law_1.2,0.15670399665832518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,32,power_law_1.2,0.16131839752197266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,32,power_law_1.01,0.1555199980735779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,32,power_law_1.2,0.15909759998321532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,32,power_law_1.2,0.16652159690856932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,32,power_law_1.01,0.15629440546035767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,32,power_law_1.2,0.1610111951828003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,32,power_law_1.2,0.17244800329208373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,32,power_law_1.01,0.6398975849151611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,32,power_law_1.01,0.15831680297851564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,32,power_law_1.2,0.17605760097503662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,32,power_law_1.2,0.16298880577087402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,32,power_law_1.01,0.16047999858856202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,32,power_law_1.01,0.801420783996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,32,power_law_1.2,0.18788479566574096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,32,power_law_1.01,0.16558079719543456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,32,power_law_1.2,0.19833600521087646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,32,power_law_1.01,0.035776001214981076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,32,power_law_1.2,0.20770559310913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,32,power_law_1.01,1.1592063903808594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,32,power_law_1.01,0.03764480054378509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,32,power_law_1.2,0.22948479652404785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,32,power_law_1.2,0.09806720018386841
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,32,power_law_1.01,0.03837440013885498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,32,power_law_1.2,0.2763967990875244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,32,power_law_1.2,0.10695040225982666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,32,power_law_1.01,1.5073216438293457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,32,power_law_1.2,0.31856000423431396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,32,power_law_1.2,0.10853760242462158
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,32,power_law_1.2,0.030368000268936157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,32,power_law_1.01,0.6910208225250244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,32,power_law_1.2,0.43080959320068357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,32,power_law_1.2,0.11290240287780762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,32,power_law_1.2,0.16825599670410157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,32,power_law_1.01,0.8836416244506836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,32,power_law_1.2,0.10831999778747559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,32,power_law_1.2,0.5419199943542481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,32,power_law_1.2,0.17330559492111205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,32,power_law_1.01,2.224870491027832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,32,power_law_1.01,0.17125760316848754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,32,power_law_1.2,0.11882879734039306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,32,power_law_1.2,0.17770240306854249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,32,power_law_1.01,0.17559679746627807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,32,power_law_1.01,1.2708736419677735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,32,power_law_1.01,0.18583680391311647
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,32,power_law_1.01,0.04632959961891174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,32,power_law_1.01,0.19634560346603394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,32,power_law_1.01,2.8680255889892576
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,32,power_law_1.01,0.08761600255966187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,32,power_law_1.01,1.5704575538635255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,32,1,balanced,0.02312533309062322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,32,1,balanced,0.023391999304294586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,32,1,balanced,0.027280000348885853
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,32,power_law_1.01,0.07701119780540466
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,32,power_law_1.2,0.028940799832344054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,32,power_law_1.01,0.06573439836502075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,32,power_law_1.2,0.028620800375938414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,32,power_law_1.01,0.05544319748878479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,32,power_law_1.01,2.361164855957031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,32,power_law_1.2,0.028531199693679808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,32,power_law_1.2,0.7269631862640381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,32,power_law_1.2,0.18806400299072265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,32,power_law_1.2,0.029356798529624938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,32,power_law_1.2,0.19962879419326782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,32,power_law_1.2,0.9269632339477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,32,power_law_1.2,0.20983679294586183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,32,power_law_1.01,0.20566399097442628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,32,power_law_1.2,0.23067519664764405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,32,1,balanced,0.043738668163617454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,32,power_law_1.01,0.22753279209136962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,32,1,balanced,0.06432533264160156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,32,power_law_1.2,0.2748863935470581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,32,1,balanced,0.09520533680915833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,32,1,balanced,0.09724266330401103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,32,power_law_1.01,0.2707520008087158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,32,1,balanced,0.09962667028109233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,32,1,balanced,0.09708799918492635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,32,power_law_1.2,0.32522881031036377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,32,1,balanced,0.09708799918492635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,32,power_law_1.01,0.3117248058319092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,32,power_law_1.2,0.1257472038269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,32,power_law_1.01,0.40242562294006345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,32,power_law_1.2,0.12889599800109863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,32,power_law_1.01,0.5268864154815673
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,32,power_law_1.2,0.030995199084281923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,32,power_law_1.01,0.6865471839904785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,32,power_law_1.2,0.03332479894161224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,32,power_law_1.01,3.110848045349121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,32,power_law_1.2,1.2574784278869628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,32,1,balanced,0.02922133356332779
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,32,power_law_1.2,0.03470079898834229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,32,power_law_1.01,0.8709823608398437
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,32,power_law_1.2,0.036556801199913024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,32,1,balanced,0.09889599680900574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,32,1,balanced,0.10148266951243083
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,32,power_law_1.2,0.03774079978466034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,32,power_law_1.2,0.4354239940643311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,32,power_law_1.01,1.2431808471679688
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,32,power_law_1.2,0.044998401403427125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,32,power_law_1.2,0.5299327850341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,32,power_law_1.2,0.13804800510406495
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,32,power_law_1.2,0.0769919991493225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,32,power_law_1.2,0.15164159536361693
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,32,power_law_1.2,0.07624959945678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,32,power_law_1.2,0.16350079774856568
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,32,power_law_1.2,0.07044479846954346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,32,power_law_1.2,0.18856960535049438
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,32,power_law_1.2,0.05978879928588867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,32,power_law_1.2,0.23754239082336426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,32,power_law_1.2,0.057529598474502563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,32,power_law_1.2,1.6923967361450196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,32,power_law_1.2,0.05556480288505554
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,32,power_law_1.2,0.05664640069007874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,32,power_law_1.2,0.06286079883575439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,32,power_law_1.01,1.5897855758666992
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,32,power_law_1.2,0.08010879755020142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,32,power_law_1.2,2.515385627746582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,32,power_law_1.2,0.09032959938049316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,32,1,balanced,0.0295413335164388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,32,1,balanced,0.03254399945338567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,32,1,balanced,0.03939199944337209
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,32,power_law_1.2,0.11949440240859985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,32,1,balanced,0.05609600245952606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,32,1,balanced,0.09291733304659526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,32,1,balanced,0.09331732988357544
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,32,power_law_1.2,0.14522240161895753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,32,power_law_1.2,0.2800704002380371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,32,1,balanced,0.10148800412813823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,32,1,balanced,0.10756267110506694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,32,1,balanced,0.1601653297742208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,32,power_law_1.2,3.292652893066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,32,1,balanced,0.16276266177495322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,32,1,balanced,0.1696000099182129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,32,1,balanced,0.18514132499694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,32,1,balanced,0.2007733384768168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,32,1,balanced,0.22980799277623495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,32,power_law_1.2,0.7085567951202393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,32,1,balanced,0.2584853370984395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,32,1,balanced,0.34165334701538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,32,1,balanced,0.4014293352762858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,32,power_law_1.2,0.9332672119140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,32,1,balanced,0.5500959952672323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,32,power_law_1.01,2.306208038330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,32,1,balanced,0.09471467137336731
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,32,power_law_1.01,0.059699201583862306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,32,power_law_1.2,1.3026944160461427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,32,power_law_1.2,0.2111423969268799
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,32,power_law_1.01,0.05420799851417542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,32,1,power_law_1.01,0.020351999998092653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,32,power_law_1.2,0.26915199756622316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,32,1,power_law_1.01,0.022438399493694305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,32,power_law_1.2,1.645529556274414
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,32,power_law_1.2,0.3749056100845337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,32,1,power_law_1.01,0.036671999096870425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,32,1,power_law_1.01,0.04417920112609863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,32,power_law_1.2,0.5194176197052002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,32,1,power_law_1.01,0.05618559718132019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,32,1,balanced,0.6934346357981364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,32,1,power_law_1.01,0.025561600923538208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,32,power_law_1.2,2.5053760528564455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,32,1,power_law_1.01,0.07039999961853027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,32,1,balanced,1.0024426778157551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,32,power_law_1.2,0.3779072046279907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,32,1,power_law_1.01,0.08113279938697815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,32,power_law_1.01,0.05780479907989502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,32,1,balanced,1.3089333375295003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,32,1,power_law_1.01,0.08223999738693237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,32,power_law_1.2,0.46862077713012695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,32,power_law_1.01,0.06209279894828797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,32,1,power_law_1.2,0.020844799280166627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,32,1,power_law_1.01,0.08556159734725952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,32,power_law_1.01,3.1383039474487306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,32,power_law_1.01,0.07495679855346679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,32,power_law_1.2,0.6586175918579101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,32,1,power_law_1.01,0.08539519906044006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,32,1,power_law_1.2,0.024512000381946564
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,32,power_law_1.01,0.08899199962615967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,32,1,power_law_1.01,0.0908415973186493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,32,1,power_law_1.2,0.027801600098609925
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,32,power_law_1.01,0.11433600187301636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,32,1,power_law_1.01,0.09462400078773499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,32,1,power_law_1.2,0.03150080144405365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,32,1,balanced,0.09409067034721375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,32,power_law_1.01,0.13979519605636598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,32,1,power_law_1.01,0.09783040285110474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,32,1,power_law_1.01,0.028601598739624024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,32,power_law_1.01,0.18782080411911012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,32,1,power_law_1.01,0.107532799243927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,32,1,power_law_1.01,0.031436800956726074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,32,1,power_law_1.01,0.15568000078201294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,32,1,power_law_1.01,0.04444159865379334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,32,1,power_law_1.01,0.15847040414810182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,32,1,power_law_1.01,0.16911360025405883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,32,1,power_law_1.2,0.02229759991168976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,32,power_law_1.2,3.369375991821289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,32,1,power_law_1.01,0.18536319732666015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,32,1,power_law_1.2,0.037273600697517395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,32,1,power_law_1.01,0.2030719995498657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,32,1,power_law_1.2,0.044121599197387694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,32,1,power_law_1.01,0.23482880592346192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,32,1,balanced,0.09548800190289815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,32,1,balanced,0.0958079993724823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,32,1,power_law_1.01,0.26847360134124754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,32,1,balanced,0.09735999504725139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,32,1,balanced,0.10116799672444661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,32,1,balanced,0.1049013336499532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,32,1,power_law_1.01,0.35710721015930175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,32,power_law_1.01,0.24193921089172363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,32,1,power_law_1.01,0.43249921798706054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,32,power_law_1.01,0.35491840839385985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,32,1,power_law_1.01,0.056857597827911374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,32,1,power_law_1.01,0.5791039943695069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,32,1,power_law_1.01,0.06738560199737549
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,32,power_law_1.2,0.849824047088623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,32,1,power_law_1.01,0.0769216001033783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,32,1,power_law_1.01,0.7362048149108886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,32,1,power_law_1.2,0.05435519814491272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,32,1,power_law_1.2,0.04392319917678833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,32,power_law_1.2,1.2132672309875487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,32,1,power_law_1.01,1.0321663856506347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,32,1,power_law_1.2,0.06737279891967773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,32,1,power_law_1.2,0.05429120063781738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,32,1,balanced,0.1482186714808146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,32,1,balanced,0.15236266454060873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,32,1,balanced,0.16058133045832315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,32,1,power_law_1.2,0.06629760265350342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,32,1,balanced,0.17734932899475098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,32,1,power_law_1.01,1.3392064094543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,32,power_law_1.2,1.5724287986755372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,32,1,power_law_1.2,0.07557759881019592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,32,1,power_law_1.2,0.07853440046310425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,32,1,power_law_1.2,0.08134400248527526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,32,1,power_law_1.2,0.08199679851531982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,32,1,power_law_1.01,0.08002560138702393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,32,1,power_law_1.2,0.08812159895896912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,32,1,power_law_1.01,0.0825984001159668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,32,1,power_law_1.2,0.09142400026321411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,32,1,power_law_1.01,0.0843455970287323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,32,1,power_law_1.2,0.08155519962310791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,32,1,power_law_1.2,0.09633920192718506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,32,1,power_law_1.01,0.08676480054855347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,32,1,power_law_1.2,0.08344320058822632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,32,1,balanced,0.19553067286809286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,32,balanced,0.0356480007370313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,32,balanced,0.0395413339138031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,32,1,power_law_1.2,0.08465279936790467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,32,balanced,0.037461332976818085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,32,balanced,0.037808001041412354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,32,balanced,0.0888266662756602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,32,1,power_law_1.2,0.0855296015739441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,32,balanced,0.1564853290716807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,32,balanced,0.16022400061289468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,32,1,power_law_1.2,0.09244800209999085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,32,power_law_1.2,2.3484352111816404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,32,1,power_law_1.2,0.10502400398254394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,32,1,balanced,0.2260800004005432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,32,1,power_law_1.2,0.14514559507369995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,32,balanced,0.03160533308982849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,32,balanced,0.03161599983771642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,32,1,power_law_1.2,0.14969600439071656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,32,balanced,0.03363200028737386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,32,balanced,0.03774933268626531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,32,balanced,0.09877333045005798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,32,balanced,0.16033066312472025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,32,balanced,0.15449600418408713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,32,balanced,0.16422399878501892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,32,balanced,0.1584213376045227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,32,balanced,0.16680532693862915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,32,balanced,0.15895467003186545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,32,balanced,0.1707680026690165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,32,balanced,0.16299733519554138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,32,balanced,0.17590934038162231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,32,balanced,0.16504533092180887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,32,balanced,0.181386669476827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,32,balanced,0.168778657913208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,32,balanced,0.19173866510391235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,32,balanced,0.17455466588338217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,32,balanced,0.20327999194463095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,32,balanced,0.21180800596872965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,32,balanced,0.23173866669336954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,32,balanced,0.2728266716003418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,32,balanced,0.3117813269297282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,32,1,balanced,0.25702933470408124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,32,balanced,0.39527467886606854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,32,power_law_1.01,0.4788991928100586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,32,1,balanced,0.35601600011189777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,32,balanced,0.4728586673736572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,32,balanced,0.048714667558670044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,32,balanced,0.05042133231957754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,32,balanced,0.6448533137639364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,32,1,power_law_1.2,0.16009600162506105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,32,1,power_law_1.2,0.09561600089073181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,32,1,power_law_1.2,0.18024959564208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,32,1,power_law_1.2,0.09966719746589661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,32,1,power_law_1.2,0.1996608018875122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,32,balanced,0.1790613333384196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,32,1,power_law_1.2,0.24003839492797852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,32,balanced,0.18968532482783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,32,balanced,0.1996799906094869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,32,1,power_law_1.2,0.2837440013885498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,32,1,power_law_1.2,0.11705600023269654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,32,1,power_law_1.01,0.09199360013008118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,32,power_law_1.2,3.172038459777832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,32,1,power_law_1.2,0.37452800273895265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,32,1,power_law_1.01,0.09594879746437072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,32,1,balanced,0.42299731572469074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,32,1,power_law_1.01,0.10598399639129638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,32,balanced,0.0506879985332489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,32,1,balanced,0.5799893140792847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,32,balanced,0.05417066812515259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,32,1,power_law_1.01,0.1446336030960083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,32,balanced,0.06444799900054932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,32,balanced,0.8019519646962484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,32,balanced,0.10934933026631673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,32,1,balanced,0.7364693482716879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,32,balanced,0.11245866616566975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,32,balanced,0.1129813293615977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,32,balanced,1.1436693668365479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,32,1,balanced,1.0538293520609539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,32,balanced,1.4656960169474285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,32,1,balanced,1.3729440371195476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,32,balanced,0.2100480000178019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,32,balanced,0.23019200563430786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,32,power_law_1.01,0.0669376015663147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,32,balanced,0.26876266797383624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,32,1,power_law_1.2,0.15601919889450072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,32,balanced,0.30795733133951825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,32,balanced,2.1282025973002114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,32,power_law_1.01,0.07136639952659607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,32,1,power_law_1.2,0.16011519432067872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,32,balanced,0.38502931594848633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,32,1,power_law_1.2,0.44611201286315916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,32,power_law_1.01,0.07123839855194092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,32,1,power_law_1.2,0.16972800493240356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,32,1,power_law_1.01,0.14890880584716798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,32,balanced,2.7874294916788735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,32,1,power_law_1.2,0.1865280032157898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,32,1,power_law_1.2,0.6071104049682617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,32,1,power_law_1.2,0.20428159236907958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,32,1,power_law_1.2,0.7677887916564942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,32,1,power_law_1.2,0.2352447986602783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,32,1,power_law_1.2,0.2719360113143921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,32,1,power_law_1.2,1.0918911933898925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,32,1,power_law_1.2,0.3617791891098022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,32,balanced,0.46013331413269043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,32,1,power_law_1.2,0.43265280723571775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,32,balanced,0.6246399879455566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,32,1,power_law_1.2,1.4282943725585937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,32,power_law_1.01,0.07148799896240235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,32,balanced,0.11582932869593303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,32,balanced,0.7746666272481283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,32,balanced,0.02757333219051361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,32,balanced,0.11940800150235494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,32,balanced,0.02741866558790207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,32,power_law_1.01,0.061260801553726194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,32,balanced,0.027450665831565857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,32,balanced,0.027514666318893433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,32,balanced,1.1231626669565837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,32,power_law_1.01,0.06190720200538635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,32,power_law_1.01,0.06207360029220581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,32,power_law_1.01,0.06322559714317322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,32,power_law_1.01,0.09303039908409119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,32,power_law_1.01,0.15251200199127196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,32,power_law_1.01,0.060185599327087405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,32,1,power_law_1.01,0.1596287965774536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,32,1,power_law_1.2,0.5814527988433837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,32,power_law_1.01,0.15635199546813966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,32,balanced,0.1232266624768575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,32,1,power_law_1.01,0.17817599773406984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,32,balanced,0.12970133622487387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,32,balanced,0.029445332785447437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,32,balanced,0.13196266690889993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,32,power_law_1.01,0.15716480016708373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,32,1,power_law_1.2,0.739404821395874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,32,1,power_law_1.01,0.19934719800949097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,32,power_law_1.01,0.160863995552063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,32,balanced,1.4299626350402832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,32,1,power_law_1.01,0.23571839332580566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,32,power_law_1.01,0.1614848017692566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,32,1,power_law_1.2,1.0407232284545898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,32,1,power_law_1.01,0.27947518825531004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,32,power_law_1.01,0.16751359701156615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,32,balanced,2.0882293383280435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,32,1,power_law_1.01,0.3677248001098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,32,power_law_1.01,0.17136000394821166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,32,1,power_law_1.2,1.3495295524597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,32,1,power_law_1.01,0.44391679763793945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,32,power_law_1.01,0.177183997631073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,32,power_law_1.01,0.06094080209732056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,32,power_law_1.01,0.10279680490493774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,32,power_law_1.01,0.18782080411911012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,32,1,power_law_1.01,0.6044735908508301
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,32,balanced,0.03362133353948593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,32,power_law_1.01,0.15055359601974488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,32,power_law_1.01,0.1964416027069092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,32,balanced,0.03691199918588003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,32,balanced,0.03782399992148081
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,32,balanced,0.03977066775163015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,32,power_law_1.01,0.20777599811553954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,32,balanced,0.04011200120051702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,32,balanced,0.050613333781560264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,32,power_law_1.01,0.2300352096557617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,32,balanced,0.10576533277829488
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,32,balanced,0.08583466211954753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,32,balanced,0.0814879983663559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,32,power_law_1.01,0.2745984077453613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,32,balanced,0.06058133145173391
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,32,balanced,0.058517331878344216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,32,balanced,0.05815466741720835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,32,balanced,0.057999998331069946
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,32,balanced,0.06046933432420095
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,32,balanced,0.06903466582298279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,32,balanced,2.772746721903483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,32,balanced,0.08164266745249431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,32,balanced,0.10521599650382996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,32,power_law_1.01,0.06635519862174988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,32,power_law_1.2,0.06769919991493226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,32,balanced,0.12596266468365988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,32,balanced,0.1716853380203247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,32,power_law_1.01,0.06652160286903382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,32,1,power_law_1.01,0.7631872177124024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,32,balanced,0.21550933519999185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,32,power_law_1.01,0.07361279726028443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,32,power_law_1.01,0.029363200068473816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,32,power_law_1.01,0.10485759973526002
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,32,power_law_1.01,0.02929919958114624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,32,power_law_1.01,0.10663039684295654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,32,power_law_1.01,0.3150592088699341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,32,power_law_1.01,0.10913920402526855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,32,power_law_1.01,0.41210241317749025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,32,power_law_1.01,0.11208959817886352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,32,balanced,0.14396799604098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,32,power_law_1.01,0.5104127883911133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,32,balanced,0.15428800384203592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,32,balanced,0.16331733266512552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,32,power_law_1.2,0.07123200297355652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,32,balanced,0.1839946707089742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,32,power_law_1.01,0.6981823921203614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,32,balanced,0.2243946592013041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,32,balanced,0.30218666791915894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,32,power_law_1.2,0.07134720087051391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,32,balanced,0.2643520037333171
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,32,balanced,0.38944534460703534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,32,balanced,0.34593598047892254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,32,1,power_law_1.01,1.0880191802978516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,32,power_law_1.2,0.0717248022556305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,32,power_law_1.01,0.8852928161621094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,32,balanced,0.4254080057144165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,32,power_law_1.2,0.061375999450683595
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,32,power_law_1.01,0.029337599873542786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,32,balanced,0.5961173375447592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,32,power_law_1.2,0.06382079720497132
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,32,power_law_1.01,0.02900480031967163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,32,power_law_1.01,1.2419648170471191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,32,1,power_law_1.01,1.413152027130127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,32,power_law_1.01,0.029900801181793214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,32,power_law_1.2,0.060192000865936277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,32,power_law_1.01,0.15389440059661866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,32,power_law_1.01,0.0310591995716095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,32,power_law_1.01,1.5981184005737306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,32,power_law_1.01,0.1564352035522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,32,power_law_1.01,0.03405439853668213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,32,power_law_1.01,0.15932159423828124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,32,power_law_1.01,0.03523840010166168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,32,power_law_1.01,0.1613759994506836
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,32,power_law_1.01,0.03844479918479919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,32,power_law_1.01,2.357721519470215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,32,power_law_1.01,0.0380160003900528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,32,balanced,0.7540586789449056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,32,power_law_1.2,0.06330239772796631
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,32,power_law_1.01,0.045151999592781066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,32,balanced,1.0763839880625408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,32,power_law_1.2,0.06381440162658691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,32,power_law_1.01,0.11446399688720703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,32,power_law_1.01,0.08101760149002075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,32,power_law_1.2,0.09178879857063293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,32,power_law_1.2,0.06204800009727478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,32,balanced,1.3965226809183757
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,32,power_law_1.01,0.07468159794807434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,32,power_law_1.01,3.1106367111206055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,32,power_law_1.2,0.0664192020893097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,32,power_law_1.2,0.1525823950767517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,32,balanced,2.034202734629313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,32,power_law_1.2,0.06631680130958557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,32,power_law_1.2,0.10065920352935791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,32,power_law_1.2,0.07171199917793274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,32,power_law_1.2,0.15129599571228028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,32,power_law_1.2,0.09893760085105896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,32,balanced,2.671173413594564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,32,1,balanced,0.023157333334287006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,32,power_law_1.2,0.1552575945854187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,32,1,balanced,0.023402666052182514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,32,power_law_1.2,0.10739840269088745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,32,1,balanced,0.02737066646416982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,32,1,balanced,0.04162133236726125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,32,1,balanced,0.06468800206979115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,32,power_law_1.2,0.15786240100860596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,32,1,balanced,0.09582400321960449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,32,power_law_1.01,0.11674879789352417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,32,1,balanced,0.097653329372406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,32,1,balanced,0.09878399968147278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,32,power_law_1.2,0.16033920049667358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,32,power_law_1.01,0.12346240282058715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,32,power_law_1.01,0.06515200138092041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,32,power_law_1.2,0.16092159748077392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,32,power_law_1.01,0.12986880540847778
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,32,power_law_1.01,0.05609599947929382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,32,power_law_1.01,0.14023679494857788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,32,power_law_1.2,0.1670016050338745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,32,power_law_1.01,0.05505920052528381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,32,power_law_1.01,0.1510655999183655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,32,power_law_1.01,0.05556480288505554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,32,power_law_1.01,0.16116479635238648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,32,power_law_1.01,0.057811200618743896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,32,power_law_1.01,0.1863935947418213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,32,power_law_1.01,0.06296319961547851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,32,power_law_1.01,0.2324671983718872
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,32,power_law_1.01,0.07737600207328796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,32,1,balanced,0.09856533010800679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,32,power_law_1.01,0.27767679691314695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,32,power_law_1.01,0.09071360230445862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,32,power_law_1.2,0.15411839485168458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,32,power_law_1.01,0.37363200187683104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,32,power_law_1.2,0.15660799741744996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,32,power_law_1.01,0.46346240043640136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,32,power_law_1.2,0.17240959405899048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,32,power_law_1.2,0.16068480014801026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,32,power_law_1.2,0.1760256052017212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,32,power_law_1.01,0.6364799976348877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,32,power_law_1.2,0.1637760043144226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,32,power_law_1.2,0.18626559972763063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,32,power_law_1.2,0.16770559549331665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,32,power_law_1.2,0.1978368043899536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,32,power_law_1.01,0.8150912284851074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,32,power_law_1.2,0.17330559492111205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,32,power_law_1.2,0.208953595161438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,32,1,balanced,0.09879466891288757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,32,1,balanced,0.09948266545931499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,32,power_law_1.2,0.1776576042175293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,32,power_law_1.2,0.23162240982055665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,32,power_law_1.01,1.1675328254699706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,32,power_law_1.2,0.18878719806671143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,32,power_law_1.2,0.2747136116027832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,32,power_law_1.01,0.166156804561615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,32,power_law_1.2,0.199891197681427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,32,power_law_1.2,0.3212032079696655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,32,power_law_1.2,0.20964479446411133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,32,power_law_1.01,1.5304896354675293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,32,power_law_1.2,0.03027839958667755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,32,power_law_1.2,0.23032960891723633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,32,power_law_1.2,0.02959359884262085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,32,power_law_1.2,0.2747391939163208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,32,power_law_1.2,0.029523199796676634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,32,power_law_1.01,2.260851287841797
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,32,power_law_1.2,0.02963840067386627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,32,1,balanced,0.10089066624641418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,32,power_law_1.01,0.11415040493011475
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,32,power_law_1.2,0.030041599273681642
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,32,power_law_1.01,0.1429759979248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,32,power_law_1.2,0.031385600566864014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,32,power_law_1.01,0.170905601978302
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,32,power_law_1.01,0.18890880346298217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,32,power_law_1.01,0.1752575993537903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,32,power_law_1.01,2.9348608016967774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,32,power_law_1.2,0.43639678955078126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,32,1,balanced,0.027610667049884796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,32,power_law_1.01,0.18600319623947142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,32,power_law_1.2,0.10529919862747192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,32,power_law_1.01,0.1956928014755249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,32,power_law_1.01,0.20747520923614501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,32,power_law_1.2,0.32947840690612795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,32,power_law_1.01,0.22913920879364014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,32,1,balanced,0.10220266381899516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,32,1,balanced,0.10897599657376607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,32,1,balanced,0.15944000085194907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,32,power_law_1.01,0.27068800926208497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,32,1,balanced,0.1630506714185079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,32,1,balanced,0.17083734273910522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,32,1,balanced,0.1851039926211039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,32,power_law_1.01,0.31405439376831057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,32,1,balanced,0.19958933194478354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,32,1,balanced,0.22823466857274374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,32,power_law_1.2,0.034560000896453856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,32,power_law_1.01,0.4186431884765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,32,1,balanced,0.2598399917284648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,32,1,balanced,0.029466666281223297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,32,power_law_1.2,0.035641598701477054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,32,1,balanced,0.03166399896144867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,32,power_law_1.01,0.5211328029632568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,32,power_law_1.2,0.5305920124053956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,32,power_law_1.2,0.03946239948272705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,32,power_law_1.2,0.11168639659881592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,32,power_law_1.01,0.689299201965332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,32,power_law_1.2,0.03937279880046844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,32,power_law_1.2,0.7175680160522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,32,power_law_1.2,0.4268928050994873
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,32,power_law_1.2,0.04643200039863586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,32,power_law_1.01,0.8824319839477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,32,power_law_1.2,0.09077759981155395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,32,power_law_1.2,0.9240511894226074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,32,power_law_1.2,0.07793920040130616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,32,power_law_1.01,0.2440959930419922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,32,power_law_1.01,1.2113471984863282
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,32,power_law_1.2,0.0591808021068573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,32,power_law_1.2,1.3018815994262696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,32,power_law_1.01,0.34311680793762206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,32,power_law_1.01,1.6040384292602539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,32,power_law_1.01,0.46392321586608887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,32,power_law_1.2,1.6960128784179687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,32,1,power_law_1.01,0.02136960029602051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,32,power_law_1.2,0.1152575969696045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,32,1,power_law_1.01,0.021971200406551362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,32,power_law_1.2,0.11989760398864746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,32,1,power_law_1.01,0.034918400645256045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,32,power_law_1.2,0.12536319494247436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,32,1,power_law_1.01,0.044012799859046936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,32,power_law_1.2,0.13025920391082763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,32,1,balanced,0.039450667798519135
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,32,power_law_1.2,0.0535103976726532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,32,power_law_1.2,0.05596799850463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,32,power_law_1.2,0.05597440004348755
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,32,power_law_1.2,0.05980799794197082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,32,power_law_1.2,0.06444799900054932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,32,power_law_1.2,0.5238336086273193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,32,power_law_1.01,2.3644735336303713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,32,power_law_1.2,0.7212287902832031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,32,power_law_1.2,2.496339225769043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,32,1,balanced,0.05834666887919108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,32,1,balanced,0.09438400467236836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,32,power_law_1.2,0.9149439811706543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,32,power_law_1.01,3.0205440521240234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,32,power_law_1.2,1.3422143936157227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,32,1,power_law_1.01,0.024255999922752382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,32,power_law_1.2,3.23559684753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,32,1,power_law_1.01,0.027910399436950683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,32,1,power_law_1.2,0.021158400177955627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,32,power_law_1.2,0.07954559922218322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,32,1,power_law_1.01,0.031430399417877196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,32,1,power_law_1.2,0.0220223993062973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,32,1,balanced,0.3404266834259033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,32,1,power_law_1.01,0.04570240080356598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,32,1,balanced,0.40186134974161786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,32,1,power_law_1.01,0.055743998289108275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,32,1,power_law_1.01,0.057004797458648684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,32,1,balanced,0.09365333120028178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,32,1,balanced,0.09392533699671428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,32,1,balanced,0.09468799829483032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,32,1,balanced,0.09453333417574565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,32,1,balanced,0.09701866904894511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,32,1,balanced,0.10011200110117595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,32,1,balanced,0.09898666540781657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,32,1,balanced,0.10326400399208069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,32,1,balanced,0.15001599987347922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,32,1,balanced,0.15236266454060873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,32,1,balanced,0.16236266493797302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,32,1,balanced,0.17837866147359213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,32,1,balanced,0.19349332650502524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,32,power_law_1.2,1.732383918762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,32,1,balanced,0.22616000970204672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,32,power_law_1.2,0.09370239973068237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,32,1,balanced,0.2588053345680237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,32,1,power_law_1.2,0.03604480028152466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,32,power_law_1.2,0.14034559726715087
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,32,power_law_1.2,0.12349439859390259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,32,1,balanced,0.5520586570103964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,32,1,power_law_1.2,0.04423039853572845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,32,1,power_law_1.01,0.07056639790534973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,32,power_law_1.2,0.14507520198822021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,32,1,power_law_1.01,0.06740480065345764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,32,1,power_law_1.2,0.05518720149993896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,32,power_law_1.2,2.556012725830078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,32,power_law_1.2,0.20100479125976561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,32,power_law_1.2,0.1504256010055542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,32,1,power_law_1.2,0.06835839748382569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,32,power_law_1.2,0.16403839588165284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,32,1,power_law_1.2,0.0812287986278534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,32,power_law_1.2,0.18836480379104614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,32,1,power_law_1.2,0.0830847978591919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,32,power_law_1.2,0.23853440284729005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,32,1,balanced,0.3553546667098999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,32,1,power_law_1.2,0.08620799779891967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,32,power_law_1.2,3.290380859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,32,power_law_1.2,0.28698880672454835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,32,1,power_law_1.2,0.08675199747085571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,32,1,balanced,0.6945333480834961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,32,power_law_1.2,0.3814847946166992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,32,1,power_law_1.01,0.08008319735527039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,32,1,balanced,1.0026826858520508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,32,power_law_1.2,0.4717567920684814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,32,1,power_law_1.01,0.08307200074195861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,32,1,power_law_1.01,0.07704960107803345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,32,1,balanced,1.3087626298268635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,32,1,power_law_1.01,0.08472319841384887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,32,1,power_law_1.01,0.07895039916038513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,32,power_law_1.2,0.6582208156585694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,32,1,power_law_1.01,0.0868287980556488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,32,1,power_law_1.01,0.08116480112075805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,32,1,power_law_1.01,0.09091839790344239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,32,power_law_1.2,0.8373632431030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,32,1,power_law_1.01,0.09438719749450683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,32,1,balanced,0.42505598068237305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,32,1,power_law_1.01,0.09829760193824769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,32,1,balanced,0.5789866844813029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,32,1,power_law_1.2,0.09265919923782348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,32,1,power_law_1.01,0.10762239694595337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,32,1,balanced,0.737178643544515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,32,1,power_law_1.2,0.09583359956741333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,32,power_law_1.2,0.2537343978881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,32,1,power_law_1.2,0.09930880069732666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,32,1,power_law_1.2,0.10794240236282349
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,32,power_law_1.2,0.39800319671630857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,32,1,power_law_1.2,0.1578112006187439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,32,1,power_law_1.01,0.08387839794158936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,32,1,power_law_1.01,0.08817279934883118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,32,1,power_law_1.2,0.024646399915218352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,32,1,power_law_1.01,0.09227520227432251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,32,power_law_1.2,1.23372802734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,32,1,power_law_1.2,0.027795198559761047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,32,1,power_law_1.01,0.09657599925994872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,32,1,power_law_1.01,0.15553280115127563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,32,1,power_law_1.2,0.035308799147605895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,32,1,power_law_1.01,0.10469759702682495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,32,1,balanced,1.056389331817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,32,1,power_law_1.01,0.1589568018913269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,32,1,power_law_1.01,0.14458240270614625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,32,1,power_law_1.2,0.044537600874900815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,32,power_law_1.2,1.6241472244262696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,32,balanced,0.035429333647092186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,32,1,power_law_1.01,0.14937599897384643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,32,balanced,0.0352960005402565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,32,1,balanced,1.3743680318196614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,32,1,power_law_1.2,0.05469440221786499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,32,balanced,0.03544000039498011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,32,balanced,0.03558400024970373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,32,balanced,0.0377866675456365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,32,balanced,0.03771200031042099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,32,1,power_law_1.2,0.066457599401474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,32,1,power_law_1.2,0.16102399826049804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,32,1,power_law_1.2,0.07649919986724854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,32,1,power_law_1.2,0.16958719491958618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,32,power_law_1.2,2.307257652282715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,32,1,power_law_1.2,0.07849599719047547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,32,1,power_law_1.2,0.18789119720458985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,32,1,power_law_1.2,0.08051199913024902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,32,1,power_law_1.2,0.20551040172576904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,32,1,power_law_1.2,0.08431360125541687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,32,1,power_law_1.2,0.23783679008483888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,32,1,power_law_1.2,0.08872320055961609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,32,1,power_law_1.01,0.16714240312576295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,32,1,power_law_1.2,0.27438080310821533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,32,power_law_1.2,2.97391357421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,32,1,power_law_1.2,0.09253759980201721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,32,1,power_law_1.01,0.1846400022506714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,32,1,power_law_1.01,0.15960960388183593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,32,1,power_law_1.2,0.3687295913696289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,32,power_law_1.01,0.05783039927482605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,32,1,power_law_1.01,0.201964807510376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,32,1,power_law_1.01,0.17893760204315184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,32,power_law_1.01,0.05750399827957153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,32,1,power_law_1.2,0.43485441207885744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,32,1,power_law_1.01,0.23408000469207763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,32,power_law_1.01,0.05848960280418396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,32,1,power_law_1.01,0.26744320392608645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,32,1,power_law_1.2,0.5853248119354248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,32,power_law_1.01,0.08135039806365967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,32,1,power_law_1.01,0.3546112060546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,32,power_law_1.01,0.0809216022491455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,32,1,power_law_1.2,0.7400256156921386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,32,power_law_1.01,0.09162240028381348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,32,1,power_law_1.01,0.43038082122802734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,32,power_law_1.01,0.09203839898109437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,32,balanced,0.07844799757003784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,32,power_law_1.01,0.09431679844856262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,32,balanced,0.07944533228874207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,32,balanced,0.09264533718427022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,32,1,power_law_1.2,0.09864320158958435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,32,balanced,0.09497599800427754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,32,balanced,0.09670933087666829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,32,balanced,0.09705600142478943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,32,balanced,0.09886933366457622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,32,1,power_law_1.01,0.19896960258483887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,32,balanced,0.10384533802668254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,32,balanced,0.10739733775456746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,32,balanced,0.10957333445549011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,32,balanced,0.11959466338157654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,32,balanced,0.12570666273434958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,32,balanced,0.1339466671148936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,32,balanced,0.15019733707110086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,32,balanced,0.18023467063903809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,32,balanced,0.20934400955835977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,32,balanced,0.2745973269144694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,32,balanced,0.3402239878972371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,32,balanced,0.47049065430959064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,32,1,power_law_1.2,1.047327995300293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,32,balanced,0.054085334142049156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,32,balanced,0.5939253171284994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,32,balanced,0.07249066730340321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,32,balanced,0.09101866682370503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,32,1,power_law_1.2,0.10597759485244751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,32,balanced,0.09091200431187947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,32,balanced,0.8686506748199463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,32,balanced,0.09539733330408733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,32,balanced,0.09609599908192952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,32,1,power_law_1.2,1.3561152458190917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,32,balanced,0.0974826713403066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,32,1,power_law_1.01,0.2361664056777954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,32,balanced,0.10212799906730652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,32,balanced,0.10572266578674316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,32,1,power_law_1.01,0.2781696081161499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,32,balanced,0.11023466785748799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,32,balanced,0.11958932876586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,32,balanced,0.12779200077056885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,32,balanced,0.13566399614016214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,32,balanced,0.15197867155075073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,32,balanced,0.18482667207717896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,32,balanced,0.2136533260345459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,32,balanced,0.277018666267395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,32,1,power_law_1.01,0.5779647827148438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,32,balanced,0.341536005338033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,32,power_law_1.01,0.09675520062446594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,32,power_law_1.2,0.5049471855163574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,32,power_law_1.01,0.09705600142478943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,32,power_law_1.2,0.05716480016708374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,32,balanced,0.47922666867574054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,32,1,power_law_1.2,0.14559359550476075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,32,1,power_law_1.01,0.7336063861846924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,32,power_law_1.2,0.05707520246505737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,32,balanced,0.6056106487909952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,32,balanced,1.1269439856211345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,32,1,power_law_1.2,0.15029120445251465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,32,power_law_1.2,0.05786240100860596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,32,power_law_1.01,0.03976320028305054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,32,balanced,0.8740906715393066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,32,power_law_1.2,0.07610880136489868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,32,1,power_law_1.01,1.0300671577453613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,32,balanced,1.6563893953959148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,32,1,power_law_1.01,0.3702847957611084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,32,power_law_1.2,0.08379520177841186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,32,balanced,1.134720007578532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,32,power_law_1.2,0.09039360284805298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,32,1,power_law_1.01,0.44394879341125487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,32,power_law_1.2,0.09093120098114013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,32,1,power_law_1.01,1.3356224060058595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,32,balanced,2.21777073542277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,32,balanced,1.6592213312784831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,32,1,balanced,0.02348266790310542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,32,power_law_1.2,0.0937279999256134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,32,1,balanced,0.02554133286078771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,32,power_law_1.2,0.04022400081157684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,32,1,balanced,0.031146667897701263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,32,power_law_1.2,0.0955456018447876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,32,power_law_1.01,0.10104320049285889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,32,balanced,2.1821866035461426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,32,power_law_1.2,0.04116480052471161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,32,1,balanced,0.025199999411900837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,32,1,balanced,0.029504001140594482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,32,1,balanced,0.035461333890755974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,32,1,power_law_1.2,0.1611199975013733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,32,1,power_law_1.2,0.18102400302886962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,32,1,power_law_1.2,0.20131199359893798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,32,1,power_law_1.2,0.24309120178222657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,32,1,power_law_1.01,0.604256010055542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,32,1,power_law_1.2,0.2845695972442627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,32,1,balanced,0.05177066723505656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,32,1,balanced,0.08521067102750142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,32,1,balanced,0.11128000418345134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,32,1,power_law_1.01,0.7657279968261719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,32,1,power_law_1.2,0.3717695951461792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,32,power_law_1.2,0.09681280255317688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,32,power_law_1.01,0.10496640205383301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,32,power_law_1.2,0.100108802318573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,32,1,power_law_1.2,0.44768638610839845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,32,power_law_1.01,0.10945919752120972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,32,1,power_law_1.01,1.0881216049194335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,32,power_law_1.01,0.03991680145263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,32,power_law_1.2,0.10559999942779541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,32,power_law_1.01,0.04041599929332733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,32,1,power_law_1.2,0.6121407985687256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,32,power_law_1.2,0.10794240236282349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,32,power_law_1.01,0.05532159805297852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,32,power_law_1.2,0.11921279430389405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,32,1,power_law_1.01,1.4123071670532226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,32,1,power_law_1.2,0.7739264011383057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,32,power_law_1.2,0.1259392023086548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,32,1,power_law_1.01,0.020927999913692475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,32,power_law_1.2,0.13422720432281493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,32,1,power_law_1.01,0.024083200097084045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,32,1,balanced,0.11302399635314941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,32,power_law_1.2,0.1510208010673523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,32,1,balanced,0.11337066690127055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,32,1,power_law_1.01,0.036934399604797365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,32,power_law_1.2,0.04071680009365082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,32,1,balanced,0.05407466491063436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,32,1,power_law_1.01,0.05691519975662231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,32,1,balanced,0.07864533364772797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,32,1,balanced,0.09705600142478943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,32,1,power_law_1.01,0.06547200083732604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,32,power_law_1.01,0.11792639493942261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,32,power_law_1.01,0.12641919851303102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,32,power_law_1.01,0.1343168020248413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,32,power_law_1.01,0.07217280268669128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,32,power_law_1.01,0.08976640105247498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,32,power_law_1.01,0.09103360176086425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,32,1,power_law_1.2,1.0941311836242675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,32,1,balanced,0.11229866743087769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,32,power_law_1.01,0.0939199984073639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,32,power_law_1.2,0.1865280032157898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,32,power_law_1.01,0.09516159892082214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,32,power_law_1.2,0.054553598165512085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,32,power_law_1.2,0.22094080448150635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,32,1,balanced,0.09708799918492635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,32,1,balanced,0.10015466809272766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,32,1,power_law_1.2,1.421504020690918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,32,power_law_1.01,0.0982208013534546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,32,1,balanced,0.09764267007509868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,32,1,balanced,0.10076799988746643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,32,1,balanced,0.10059733192125957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,32,power_law_1.01,0.10205440521240235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,32,1,balanced,0.10630399982134502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,32,1,balanced,0.10987200339635213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,32,power_law_1.01,0.1059648036956787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,32,power_law_1.01,0.10813440084457397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,32,power_law_1.01,0.11850880384445191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,32,power_law_1.01,0.12750719785690307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,32,1,balanced,0.1126026709874471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,32,1,balanced,0.1132586697737376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,32,power_law_1.01,0.13426560163497925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,32,power_law_1.01,0.15135999917984008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,32,power_law_1.2,0.07232000231742859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,32,power_law_1.2,0.3032383918762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,32,power_law_1.01,0.1905727982521057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,32,power_law_1.2,0.08967040181159973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,32,power_law_1.2,0.4006336212158203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,32,power_law_1.01,0.21962881088256836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,32,1,balanced,0.11045866211255391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,32,1,balanced,0.11125333110491435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,32,power_law_1.01,0.2994688034057617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,32,1,balanced,0.11748799681663513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,32,1,balanced,0.12426132957140605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,32,1,balanced,0.14326933026313782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,32,1,power_law_1.01,0.025574401021003723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,32,1,power_law_1.01,0.027449598908424376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,32,1,balanced,0.13014400005340576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,32,1,balanced,0.1306880017121633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,32,1,power_law_1.01,0.034745600819587705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,32,1,power_law_1.01,0.08891519904136658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,32,1,power_law_1.01,0.09362559914588928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,32,1,power_law_1.01,0.09690240025520325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,32,power_law_1.2,0.09104639887809754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,32,1,power_law_1.01,0.10092799663543701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,32,power_law_1.2,0.09358720183372497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,32,power_law_1.2,0.5780416011810303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,32,1,power_law_1.01,0.10592639446258545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,32,power_law_1.2,0.09452160000801087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,32,1,power_law_1.01,0.11123199462890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,32,power_law_1.2,0.09660800099372864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,32,power_law_1.2,0.7411392211914063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,32,1,balanced,0.15688000122706094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,32,1,power_law_1.01,0.12412799596786499
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,32,power_law_1.2,0.10028799772262573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,32,1,balanced,0.21890133619308472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,32,1,power_law_1.01,0.12454400062561036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,32,power_law_1.2,0.1044160008430481
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,32,1,power_law_1.01,0.12896640300750734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,32,power_law_1.2,1.0757184028625488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,32,power_law_1.2,0.10900479555130005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,32,1,power_law_1.01,0.13396480083465576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,32,power_law_1.2,0.11868159770965576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,32,1,power_law_1.01,0.13957120180130006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,32,power_law_1.2,0.12602880001068115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,32,power_law_1.2,1.4945599555969238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,32,power_law_1.01,0.15066239833831788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,32,power_law_1.2,0.1356608033180237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,32,power_law_1.01,0.18376319408416747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,32,power_law_1.2,0.15279359817504884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,32,1,power_law_1.01,0.1491328001022339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,32,power_law_1.01,0.380729603767395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,32,power_law_1.2,0.18849920034408568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,32,1,power_law_1.01,0.16728320121765136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,32,1,power_law_1.01,0.04997760057449341
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,32,1,power_law_1.01,0.1843135952949524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,32,power_law_1.2,0.2292479991912842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,32,1,power_law_1.01,0.06042240262031555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,32,1,balanced,0.24238399664560953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,32,power_law_1.01,0.5493696212768555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,32,1,power_law_1.01,0.22072958946228027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,32,1,power_law_1.01,0.08327040076255798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,32,power_law_1.2,0.3060992002487183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,32,1,power_law_1.01,0.25564160346984866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,32,power_law_1.01,0.7044032096862793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,32,power_law_1.2,0.383955192565918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,32,1,power_law_1.01,0.3289599895477295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,32,power_law_1.2,0.5814527988433837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,32,1,power_law_1.01,0.41291518211364747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,32,1,power_law_1.01,0.5774911880493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,32,power_law_1.2,0.7333312034606934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,32,power_law_1.01,0.2176896095275879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,32,power_law_1.01,0.28709120750427247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,32,1,power_law_1.01,0.7305344104766845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,32,power_law_1.2,1.0715007781982422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,32,1,balanced,0.33990931510925293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,32,1,balanced,0.38849600156148273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,32,1,power_law_1.01,1.0467391967773438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,32,1,balanced,0.5715039968490601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,32,power_law_1.2,1.4038528442382812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,32,1,balanced,0.7116906642913818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,32,power_law_1.01,0.9958335876464843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,32,1,power_law_1.01,1.3593728065490722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,32,1,balanced,1.031920035680135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,32,1,power_law_1.2,0.020640000700950623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,32,1,power_law_1.2,0.024243199825286867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,32,1,balanced,1.3543413480122883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,32,power_law_1.01,1.3537792205810546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,32,1,power_law_1.2,0.03628160059452057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,32,power_law_1.2,2.135539245605469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,32,1,power_law_1.2,0.052102398872375486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,32,1,power_law_1.2,0.026015999913215637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,32,1,balanced,0.13582932949066162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,32,1,power_law_1.2,0.06223999857902527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,32,1,balanced,0.1381439963976542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,32,1,power_law_1.01,0.08743680119514466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,32,1,power_law_1.2,0.02746880054473877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,32,1,balanced,0.13992533087730408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,32,1,power_law_1.2,0.08923519849777221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,32,1,power_law_1.01,0.08985599875450134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,32,1,power_law_1.2,0.03495039939880371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,32,1,power_law_1.2,0.09670400023460388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,32,1,power_law_1.01,0.0945088028907776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,32,1,power_law_1.2,0.04783360064029694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,32,1,power_law_1.2,0.09747840166091919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,32,1,power_law_1.01,0.0961023986339569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,32,1,power_law_1.2,0.05894399881362915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,32,1,power_law_1.01,0.1054527997970581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,32,1,power_law_1.2,0.0810368001461029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,32,1,power_law_1.2,0.08760960102081299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,32,1,power_law_1.2,0.08826239705085755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,32,1,power_law_1.2,0.09316480159759521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,32,power_law_1.01,0.375929594039917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,32,1,power_law_1.2,0.09675520062446594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,32,1,balanced,0.14602133631706238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,32,1,power_law_1.2,0.10815999507904053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,32,1,balanced,0.15823466579119363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,32,power_law_1.01,1.9827072143554687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,32,1,power_law_1.2,0.10359679460525513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,32,1,power_law_1.2,0.10587519407272339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,32,1,power_law_1.2,0.10224640369415283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,32,1,power_law_1.2,0.10689280033111573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,32,1,power_law_1.2,0.10368000268936158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,32,1,power_law_1.2,0.11474560499191284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,32,1,power_law_1.2,0.11354240179061889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,32,1,power_law_1.2,0.11758079528808593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,32,power_law_1.01,2.698067283630371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,32,1,power_law_1.2,0.1318336009979248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,32,balanced,0.043866669138272606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,32,balanced,0.07924800117810567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,32,balanced,0.09724799791971843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,32,balanced,0.16761600971221924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,32,balanced,0.2423306703567505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,32,power_law_1.01,0.5475200176239013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,32,balanced,0.5214506785074869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,32,1,balanced,0.16910399993260702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,32,1,balanced,0.22984000047047934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,32,balanced,0.5325599908828735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,32,power_law_1.01,0.7047296047210694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,32,balanced,0.5403519868850708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,32,balanced,0.5490933259328207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,32,power_law_1.01,1.040396785736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,32,1,power_law_1.2,0.12369920015335083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,32,1,power_law_1.01,0.10349440574645996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,32,1,power_law_1.2,0.12730239629745482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,32,power_law_1.01,1.2947263717651367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,32,1,power_law_1.01,0.10618239641189575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,32,1,power_law_1.2,0.13178880214691163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,32,1,power_law_1.01,0.11085439920425415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,32,1,power_law_1.2,0.13672959804534912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,32,power_law_1.2,2.2017919540405275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,32,1,power_law_1.01,0.11532800197601319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,32,1,power_law_1.2,0.1408128023147583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,32,1,power_law_1.01,0.1198848009109497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,32,1,balanced,0.2512800097465515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,32,1,power_law_1.2,0.14911999702453613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,32,1,power_law_1.01,0.1276479959487915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,32,1,balanced,0.3350133498509725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,32,1,power_law_1.2,0.16951040029525757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,32,1,balanced,0.37776533762613934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,32,1,power_law_1.01,0.15244799852371216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,32,1,power_law_1.2,0.1865023970603943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,32,1,balanced,0.570522665977478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,32,1,power_law_1.01,0.1723904013633728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,32,1,power_law_1.2,0.22307200431823732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,32,1,balanced,0.6987573305765787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,32,1,power_law_1.01,0.2124864101409912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,32,1,power_law_1.2,0.2593024015426636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,32,1,power_law_1.01,0.2492288112640381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,32,1,balanced,1.0105653603871663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,32,1,power_law_1.2,0.33182079792022706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,32,1,power_law_1.01,0.33435521125793455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,32,1,balanced,1.327455997467041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,32,power_law_1.2,2.9319103240966795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,32,1,power_law_1.2,0.4148096084594727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,32,balanced,0.045925334095954895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,32,balanced,0.07321066657702129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,32,balanced,0.08228800197442372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,32,power_law_1.01,2.019366455078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,32,balanced,0.12940800189971924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,32,1,power_law_1.2,0.5790783882141113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,32,balanced,0.18569066127141318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,32,balanced,0.4906933307647705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,32,balanced,0.5584319829940796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,32,balanced,0.49812265237172443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,32,balanced,0.5760800043741862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,32,balanced,0.5946613152821859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,32,power_law_1.01,2.668614387512207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,32,balanced,0.6138026714324951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,32,balanced,0.6508906682332357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,32,power_law_1.01,0.23402879238128663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,32,balanced,0.6849066416422526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,32,1,power_law_1.01,0.41805438995361327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,32,balanced,0.7256266276041666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,32,balanced,0.06645333270231883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,32,power_law_1.01,0.22799999713897706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,32,balanced,0.08191466828187306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,32,balanced,0.798805316289266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,32,1,power_law_1.01,0.5792640209197998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,32,balanced,0.09153599540392558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,32,balanced,0.15799466768900552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,32,balanced,0.9498026371002197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,32,balanced,0.25729600588480633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,32,power_law_1.01,0.22791039943695068
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,32,1,power_law_1.01,0.7391039848327636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,32,balanced,1.0964159965515137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,32,power_law_1.01,0.22799999713897706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,32,1,power_law_1.2,0.1552191972732544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,32,1,power_law_1.01,1.0656831741333008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,32,balanced,1.450160026550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,32,1,power_law_1.2,0.17490559816360474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,32,power_law_1.01,0.21568000316619873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,32,1,power_law_1.2,0.21413118839263917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,32,balanced,1.7293920516967773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,32,1,power_law_1.2,0.2507456064224243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,32,power_law_1.01,0.48368000984191895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,32,1,power_law_1.2,0.33656959533691405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,32,1,power_law_1.2,0.732915210723877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,32,balanced,2.426128069559733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,32,balanced,0.32256533702214557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,32,power_law_1.01,0.5108863830566406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,32,balanced,0.3326080044110616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,32,balanced,0.34277331829071045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,32,1,power_law_1.2,1.0521984100341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,32,power_law_1.01,0.5161600112915039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,32,balanced,3.056912104288737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,32,power_law_1.01,0.5182911872863769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,32,1,power_law_1.01,1.3899711608886718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,32,power_law_1.01,0.545036792755127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,32,1,power_law_1.2,1.3747008323669434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,32,power_law_1.01,0.15590399503707886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,32,power_law_1.01,0.5706624031066895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,32,balanced,4.473466555277507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,32,power_law_1.01,0.5818943977355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,32,power_law_1.01,0.15319039821624755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,32,balanced,0.35117868582407635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,32,balanced,0.3611146608988444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,32,power_law_1.01,0.5985536098480224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,32,power_law_1.01,0.155731201171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,32,1,power_law_1.2,0.4192255973815918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,32,power_law_1.01,0.6396031856536866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,32,power_law_1.01,0.15845760107040405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,32,1,power_law_1.2,0.5805759906768799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,32,balanced,5.808858871459961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,32,power_law_1.01,0.693612813949585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,32,power_law_1.01,0.1867840051651001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,32,1,power_law_1.2,0.744652795791626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,32,balanced,0.5083413521448771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,32,power_law_1.01,0.7527616024017334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,32,power_law_1.01,0.25502719879150393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,32,power_law_1.01,0.15682560205459595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,32,1,power_law_1.2,1.0712960243225098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,32,power_law_1.01,0.8277695655822754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,32,power_law_1.01,0.3156928062438965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,32,power_law_1.01,0.15779839754104613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,32,balanced,0.37841065724690753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,32,balanced,0.39838401476542157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,32,power_law_1.01,0.3162559986114502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,32,power_law_1.01,0.1559231996536255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,32,1,power_law_1.2,1.3972031593322753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,32,balanced,0.420906662940979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,32,balanced,0.4549119869867961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,32,power_law_1.01,0.333024001121521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,32,power_law_1.01,0.17509119510650634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,32,balanced,0.4904160102208455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,32,power_law_1.2,0.23432960510253906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,32,balanced,8.589925130208334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,32,power_law_1.01,0.3478528022766113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,32,power_law_1.01,0.47852158546447754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,32,power_law_1.2,0.2396928071975708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,32,balanced,0.5173706610997518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,32,power_law_1.01,0.35514240264892577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,32,balanced,0.527071992556254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,32,power_law_1.01,0.4814015865325928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,32,balanced,0.5457386573155721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,32,power_law_1.2,0.23995521068572997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,32,balanced,0.5634986559549967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,32,power_law_1.01,0.49240961074829104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,32,balanced,0.5871520042419434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,32,power_law_1.2,0.24119679927825927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,32,power_law_1.01,0.9844032287597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,32,balanced,0.6247573296229044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,32,power_law_1.01,0.49767041206359863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,32,balanced,0.6603999932607015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,32,power_law_1.2,0.23566079139709473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,32,power_law_1.01,1.1842944145202636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,32,power_law_1.01,0.5157695770263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,32,balanced,0.5303253332773844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,32,balanced,0.6044693390528361
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,32,power_law_1.01,0.5364031791687012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,32,power_law_1.2,0.4697216033935547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,32,balanced,0.7486986319224039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,32,power_law_1.01,0.3759680032730103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,32,balanced,0.8931519985198975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,32,power_law_1.2,0.48288640975952146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,32,balanced,11.259051005045572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,32,power_law_1.01,0.3921152114868164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,32,balanced,1.2476853529612224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,32,power_law_1.2,0.5217599868774414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,32,power_law_1.2,0.15393279790878295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,32,power_law_1.01,0.4157951831817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,32,balanced,1.531434694925944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,32,power_law_1.2,0.5059135913848877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,32,balanced,0.7013066609700521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,32,power_law_1.2,0.15614080429077148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,32,power_law_1.01,0.45058560371398926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,32,balanced,0.7760213216145834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,32,power_law_1.2,0.5264512062072754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,32,power_law_1.2,0.15713920593261718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,32,power_law_1.01,0.4926464080810547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,32,balanced,2.159701347351074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,32,balanced,0.9260000387827555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,32,power_law_1.2,0.5719679832458496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,32,power_law_1.01,0.5560959815979004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,32,power_law_1.2,0.15830399990081787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,32,balanced,1.0730079809824626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,32,power_law_1.2,0.579756784439087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,32,power_law_1.01,0.5787775993347168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,32,balanced,1.4406612714131672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,32,power_law_1.2,0.6096447944641114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,32,power_law_1.01,0.6215616226196289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,32,balanced,1.730138619740804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,32,power_law_1.2,0.6699840068817139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,32,power_law_1.01,0.6723584175109864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,32,power_law_1.2,0.7207680225372315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,32,power_law_1.01,0.7181056022644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,32,power_law_1.2,2.9672191619873045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,32,power_law_1.01,0.545849609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,32,power_law_1.2,0.7520895957946777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,32,power_law_1.01,0.6277567863464355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,32,power_law_1.2,0.15241600275039674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,32,power_law_1.2,0.8364800453186035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,32,power_law_1.01,0.7924672126770019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,32,power_law_1.2,0.15514880418777466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,32,power_law_1.01,0.9656384468078614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,32,power_law_1.2,0.15809279680252075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,32,balanced,2.39737606048584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,32,power_law_1.01,1.5261568069458007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,32,power_law_1.2,0.18652160167694093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,32,power_law_1.01,1.3130047798156739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,32,power_law_1.01,0.8127872467041015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,32,power_law_1.2,0.1655743956565857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,32,balanced,2.9860639572143555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,32,power_law_1.2,0.2552128076553345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,32,power_law_1.01,0.9958911895751953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,32,power_law_1.01,1.8842432022094726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,32,power_law_1.01,1.640083122253418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,32,power_law_1.2,0.3072000026702881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,32,power_law_1.01,1.1701696395874024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,32,power_law_1.2,1.0364928245544434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,32,power_law_1.2,0.31112959384918215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,32,balanced,4.4211680094401045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,32,power_law_1.01,2.558323287963867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,32,power_law_1.2,0.32539520263671873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,32,power_law_1.01,1.5559743881225585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,32,power_law_1.2,1.1977151870727538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,32,power_law_1.2,0.343340802192688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,32,balanced,2.7242186864217124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,32,power_law_1.2,0.35484158992767334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,32,power_law_1.2,1.6200639724731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,32,power_law_1.01,1.8685184478759767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,32,power_law_1.2,0.37049601078033445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,32,power_law_1.01,3.2866752624511717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,32,power_law_1.2,0.39412479400634765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,32,balanced,3.9882399241129556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,32,power_law_1.2,1.9153535842895508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,32,power_law_1.2,0.4188992023468018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,32,power_law_1.01,2.3433727264404296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,32,power_law_1.01,2.6297088623046876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,32,power_law_1.2,0.45864319801330566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,32,power_law_1.2,0.5048319816589355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,32,power_law_1.2,2.6883264541625977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,32,power_law_1.2,0.47849597930908205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,32,balanced,5.18394120534261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,32,power_law_1.01,4.756582260131836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,32,balanced,5.7804107666015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,32,power_law_1.01,2.997235107421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,32,power_law_1.2,0.5503168106079102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,32,power_law_1.01,3.358835220336914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,32,power_law_1.2,0.4804863929748535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,32,power_law_1.2,0.6442048072814941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,32,power_law_1.2,0.5024960041046143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,32,power_law_1.2,3.464857482910156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,32,power_law_1.2,0.49831042289733884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,32,power_law_1.2,0.8146559715270996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,32,power_law_1.2,0.5189375877380371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,32,power_law_1.01,4.386323165893555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,32,power_law_1.2,0.9725567817687988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,32,power_law_1.2,0.5381951808929444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,32,power_law_1.01,4.810489654541016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,32,power_law_1.01,6.204019165039062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,32,power_law_1.2,0.5611648082733154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,32,balanced,8.604645411173502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,32,power_law_1.2,1.3628864288330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,32,power_law_1.2,0.589299201965332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,32,power_law_1.2,5.102048110961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,32,power_law_1.2,0.6392640113830567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,32,power_law_1.2,1.679110336303711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,32,power_law_1.2,0.6892735958099365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,32,balanced,7.666981379191081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,32,power_law_1.01,5.6633857727050785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,32,power_law_1.2,0.7279168128967285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,32,power_law_1.01,6.372723388671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,32,power_law_1.2,2.435468864440918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,32,power_law_1.2,0.8216959953308105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,32,balanced,11.254693349202475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,32,power_law_1.2,1.009017562866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,32,power_law_1.2,7.284684753417968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,32,power_law_1.2,3.220800018310547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,32,power_law_1.2,1.2033663749694825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,32,balanced,10.065850575764975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,32,power_law_1.2,1.5960448265075684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,32,power_law_1.01,8.455833435058594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,32,power_law_1.01,9.287564849853515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,32,power_law_1.2,1.967430305480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,32,power_law_1.2,4.594291305541992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,32,power_law_1.01,9.359481811523438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,32,1,balanced,0.04219200213750204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,32,power_law_1.2,2.6632831573486326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,32,power_law_1.2,10.211532592773438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,32,1,balanced,0.04901866614818573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,32,1,balanced,0.07482133309046428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,32,1,balanced,0.11343999703725179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,32,power_law_1.2,5.962809753417969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,32,power_law_1.2,3.51214714050293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,32,1,balanced,0.07444799939791362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,32,1,balanced,0.11981866757074992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,32,power_law_1.01,11.258726501464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,32,power_law_1.01,12.848127746582032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,32,power_law_1.2,5.165599822998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,32,power_law_1.01,12.540499114990235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,32,1,power_law_1.01,0.04624640047550201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,32,1,power_law_1.2,0.04049279987812042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,32,1,power_law_1.01,0.06347519755363465
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,32,1,power_law_1.2,0.07514240145683289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,32,1,balanced,0.2265546719233195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,32,1,balanced,0.41490666071573895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,32,1,power_law_1.01,0.11302399635314941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,32,1,power_law_1.2,0.1177664041519165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,32,1,balanced,0.5890719890594482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,32,1,balanced,0.5955893198649088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,32,1,power_law_1.01,0.20648319721221925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,32,1,balanced,0.2400373419125875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,32,1,power_law_1.2,0.19950720071792602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,32,1,balanced,0.44284268220265705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,32,1,balanced,0.6481066544850668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,32,1,power_law_1.01,0.289516806602478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,32,power_law_1.2,6.641043090820313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,32,1,power_law_1.2,0.28244481086730955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,32,1,balanced,0.6513493458429972
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,32,1,power_law_1.01,0.040217599272727965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,32,1,balanced,0.6577386856079102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,32,1,power_law_1.01,0.46523518562316896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,32,1,power_law_1.2,0.488262414932251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,32,1,power_law_1.01,0.07200000286102295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,32,1,power_law_1.01,0.4953919887542725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,32,1,power_law_1.01,0.11678080558776856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,32,1,power_law_1.01,0.5086336135864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,32,1,balanced,0.5997706651687622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,32,1,balanced,0.6024693250656128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,32,power_law_1.2,15.194905090332032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,32,1,power_law_1.01,0.5260672092437744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,32,1,balanced,0.6077653169631958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,32,1,balanced,0.6161706844965616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,32,1,power_law_1.01,0.547430419921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,32,1,balanced,0.6275146802266439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,32,1,balanced,0.6600746711095175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,32,1,balanced,0.6401013135910034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,32,1,power_law_1.01,0.6033599853515625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,32,1,balanced,0.660869320233663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,32,1,power_law_1.2,0.5226431846618652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,32,1,power_law_1.01,0.6506624221801758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,32,power_law_1.2,9.916422271728516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,32,1,power_law_1.01,0.7053247928619385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,32,1,power_law_1.01,0.8037887573242187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,32,1,power_law_1.01,0.21775999069213867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,32,1,power_law_1.01,0.8983039855957031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,32,1,balanced,0.6611146529515585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,32,1,power_law_1.01,0.3101311922073364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,32,1,power_law_1.2,0.04570879936218262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,32,1,balanced,0.6771466732025146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,32,1,power_law_1.01,0.9933504104614258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,32,1,balanced,0.9367626508076986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,32,1,power_law_1.2,0.5420351982116699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,32,1,power_law_1.01,0.4912576198577881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,32,1,power_law_1.01,1.1967040061950684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,32,1,balanced,1.224303960800171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,32,1,power_law_1.2,0.5791296005249024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,32,1,power_law_1.01,0.5268799781799316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,32,1,power_law_1.01,0.8185728073120118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,32,1,power_law_1.2,0.5997568130493164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,32,1,power_law_1.01,0.5504000186920166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,32,1,power_law_1.01,0.9293503761291504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,32,1,power_law_1.01,0.5760575771331787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,32,1,balanced,0.6707733472188314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,32,1,balanced,0.6757919788360596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,32,1,power_law_1.01,1.1550399780273437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,32,1,power_law_1.01,0.581171178817749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,32,1,power_law_1.2,0.07496960163116455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,32,1,balanced,0.688368002573649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,32,1,balanced,0.6983093420664469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,32,1,power_law_1.01,0.6362751960754395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,32,1,power_law_1.2,0.11232000589370728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,32,1,power_law_1.01,1.3696895599365235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,32,1,balanced,0.7110346953074137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,32,1,balanced,0.9750346342722574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,32,1,power_law_1.01,0.6839360237121582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,32,1,balanced,1.2670986652374268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,32,1,power_law_1.01,1.8150911331176758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,32,1,power_law_1.01,0.7397312164306641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,32,1,power_law_1.01,0.8292991638183593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,32,1,power_law_1.01,2.250195121765137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,32,1,power_law_1.01,0.9277759552001953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,32,1,power_law_1.01,1.037484836578369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,32,1,power_law_1.2,0.6493824005126954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,32,1,balanced,0.7362720171610514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,32,1,power_law_1.01,3.140870475769043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,32,1,power_law_1.01,1.2377344131469727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,32,1,power_law_1.2,0.69966721534729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,32,1,balanced,0.7931093374888102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,32,1,power_law_1.2,0.1924288034439087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,32,1,balanced,1.245791991551717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,32,1,power_law_1.2,0.7582399845123291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,32,1,power_law_1.2,0.2587968111038208
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,32,1,balanced,1.3503732681274414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,32,1,power_law_1.2,0.863372802734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,32,1,power_law_1.2,0.45450239181518554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,32,1,power_law_1.01,4.030060958862305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,32,1,balanced,1.902890682220459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,32,1,power_law_1.2,0.9660544395446777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,32,1,power_law_1.2,0.48952322006225585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,32,1,balanced,2.1230239868164062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,32,1,power_law_1.2,1.0683135986328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,32,1,balanced,0.7760480244954427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,32,1,balanced,0.8257760206858317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,32,1,power_law_1.2,1.265017604827881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,32,1,balanced,3.2142133712768555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,32,1,balanced,1.26800537109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,32,1,power_law_1.2,0.8595392227172851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,32,1,balanced,1.364240010579427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,32,1,power_law_1.2,0.5021759986877441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,32,1,power_law_1.2,0.9689408302307129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,32,1,balanced,1.935765266418457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,32,1,power_law_1.2,0.5299327850341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,32,1,power_law_1.2,1.1918399810791016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,32,1,balanced,2.1347254117329917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,32,1,power_law_1.2,1.407583999633789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,32,1,power_law_1.01,5.876838302612304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,32,1,power_law_1.01,0.8330047607421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,32,1,balanced,3.2515999476114907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,32,1,power_law_1.01,0.9498047828674316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,32,1,balanced,3.9959465662638345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,32,1,power_law_1.2,1.8492992401123047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,32,power_law_1.2,13.695475769042968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,32,1,power_law_1.01,1.1740863800048829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,64,balanced,0.07650133470694225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,64,balanced,0.0766186664501826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,64,balanced,0.0775626649459203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,64,balanced,0.1406773328781128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,64,balanced,0.19569599628448486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,32,1,power_law_1.2,2.295871925354004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,64,balanced,0.27450132369995117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,32,1,power_law_1.01,1.3832256317138671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,64,balanced,0.2789813280105591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,64,balanced,0.2821333408355713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,64,balanced,0.28652799129486084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,32,1,balanced,5.917845408121745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,64,balanced,0.29055466254552204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,64,balanced,0.29609066247940063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,64,balanced,0.30269332726796466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,32,1,power_law_1.01,7.689568328857422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,64,balanced,0.05003733436266581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,32,1,power_law_1.2,3.1684032440185548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,64,balanced,0.04596266647179922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,64,balanced,0.050111999114354454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,64,balanced,0.10123200217882793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,32,1,power_law_1.2,0.5600128173828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,64,balanced,0.16331199804941812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,64,balanced,0.26255999008814496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,64,balanced,0.26846933364868164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,32,1,power_law_1.2,0.6094207763671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,64,balanced,0.2710826595624288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,64,balanced,0.2751680016517639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,64,balanced,0.27937066555023193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,32,1,power_law_1.2,0.6674623966217041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,64,balanced,0.2860693335533142
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,64,balanced,0.2972533305486043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,64,balanced,0.303930660088857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,32,1,power_law_1.2,4.06242561340332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,64,balanced,0.31824533144632977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,64,balanced,0.3370026747385661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,64,balanced,0.3532053232192993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,64,balanced,0.3871413469314575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,32,1,balanced,4.028826713562012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,64,balanced,0.44923198223114014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,64,balanced,0.5108693440755209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,64,balanced,0.639520009358724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,64,balanced,0.7657813231150309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,32,power_law_1.2,8.793036651611327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,32,1,balanced,7.523509343465169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,64,balanced,0.07391466697057088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,32,1,power_law_1.2,0.7272768020629883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,64,balanced,0.07425599793593089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,64,balanced,0.08277333279450734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,32,1,power_law_1.01,1.8254335403442383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,32,1,balanced,5.953818639119466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,32,1,power_law_1.2,5.897849655151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,32,1,power_law_1.2,0.8324735641479493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,32,1,power_law_1.01,2.2646656036376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,64,balanced,1.0348479747772217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,64,balanced,1.2848532994588215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,64,balanced,0.3099946578343709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,32,1,balanced,7.505280176798503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,64,balanced,0.32202666997909546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,32,1,power_law_1.01,3.1515392303466796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,64,balanced,0.33638401826222736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,64,balanced,0.3497759898503621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,64,balanced,1.81876802444458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,64,balanced,0.37990931669871014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,64,balanced,0.44017601013183594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,32,1,power_law_1.2,0.9281536102294922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,64,balanced,0.5021493434906006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,64,balanced,2.3635573387145996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,64,balanced,0.625216007232666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,32,1,power_law_1.2,1.023193645477295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,32,1,power_law_1.01,4.043545532226562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,64,balanced,0.08105066418647766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,64,balanced,3.4294986724853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,64,balanced,0.11184533437093098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,32,1,power_law_1.2,1.2347647666931152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,64,balanced,0.1688693364461263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,64,balanced,0.17090133825937906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,64,balanced,0.1732906699180603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,64,balanced,0.18028799692789713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,64,balanced,0.18239466349283853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,64,balanced,0.19135467211405435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,64,balanced,0.1972800095876058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,64,balanced,0.20601600408554077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,64,balanced,4.51415475209554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,64,balanced,0.2189226746559143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,64,balanced,0.7454826831817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,64,balanced,0.2360693415006002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,64,power_law_1.01,0.1062656044960022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,64,balanced,0.2515146732330322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,32,1,power_law_1.01,5.87589111328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,64,power_law_1.01,0.10767999887466431
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,64,power_law_1.01,0.10717439651489258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,64,power_law_1.01,0.1652799963951111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,32,1,power_law_1.2,7.702444458007813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,64,power_law_1.01,0.19864959716796876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,64,power_law_1.01,0.09313279986381531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,64,power_law_1.01,0.2720896005630493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,64,balanced,0.029701332251230877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,64,balanced,0.03145066648721695
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,64,balanced,0.03147733211517334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,64,balanced,0.03154666721820831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,64,balanced,0.03381866713364919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,64,balanced,0.04589866598447164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,64,balanced,0.06237333516279856
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,64,balanced,0.05131733417510986
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,64,balanced,0.13455466429392496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,64,balanced,0.13362666964530945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,64,balanced,0.12599999705950418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,32,1,power_law_1.01,7.667359924316406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,64,power_law_1.01,0.08532480001449586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,64,power_law_1.01,0.09067519903182983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,64,power_law_1.01,0.09269760251045227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,64,power_law_1.01,0.27520639896392823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,64,power_law_1.01,0.09992960095405579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,64,balanced,1.0043093363444011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,64,power_law_1.01,0.09401599764823913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,64,power_law_1.01,0.09941120147705078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,64,balanced,1.2495306332906086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,64,power_law_1.01,0.1363968014717102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,64,power_law_1.01,0.11509759426116943
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,64,balanced,0.11080533266067505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,64,power_law_1.01,0.17021440267562865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,64,balanced,1.7733012835184734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,64,power_law_1.01,0.2608511924743652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,64,power_law_1.01,0.26492159366607665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,64,balanced,2.310160001118978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,32,1,power_law_1.2,0.8302016258239746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,64,balanced,0.2855733235677083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,64,balanced,0.34374932448069256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,64,power_law_1.01,0.27946879863739016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,64,balanced,0.4026240110397339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,32,1,power_law_1.2,0.9474880218505859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,64,balanced,3.4032160441080728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,64,power_law_1.01,0.28371200561523435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,64,power_law_1.01,0.15834879875183105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,64,power_law_1.01,0.2876032114028931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,32,1,power_law_1.2,1.1734784126281739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,64,power_law_1.01,0.16891520023345946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,64,power_law_1.01,0.2934015989303589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,64,power_law_1.01,0.17203199863433838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,64,power_law_1.01,0.2997888088226318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,32,1,power_law_1.2,1.3879615783691406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,64,balanced,4.6159413655598955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,64,power_law_1.01,0.1753216028213501
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,64,power_law_1.01,0.30755839347839353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,64,power_law_1.01,0.17945599555969238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,32,power_law_1.2,11.898611450195313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,32,1,power_law_1.2,1.8307584762573241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,64,balanced,0.5226826667785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,64,power_law_1.2,0.10533759593963624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,64,balanced,0.6423360109329224
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,64,balanced,0.09872532884279887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,64,power_law_1.2,0.10526080131530761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,64,balanced,0.9026453495025635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,64,power_law_1.2,0.10559999942779541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,32,1,power_law_1.2,2.2678848266601563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,64,balanced,1.1393653551737468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,64,power_law_1.2,0.16346880197525024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,64,power_law_1.01,0.2669503927230835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,64,power_law_1.2,0.19717760086059571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,64,balanced,1.6258079210917156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,64,power_law_1.01,0.272652792930603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,64,power_law_1.01,0.318342399597168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,64,power_law_1.01,0.18740479946136473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,64,power_law_1.01,0.27628800868988035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,64,power_law_1.01,0.3341952085494995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,32,1,power_law_1.2,3.1556224822998047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,64,balanced,2.126293341318766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,64,power_law_1.01,0.19668480157852172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,64,power_law_1.01,0.34755840301513674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,64,balanced,0.0765226682027181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,64,power_law_1.01,0.2041408061981201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,64,power_law_1.01,0.3806207895278931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,64,power_law_1.01,0.21706879138946533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,64,power_law_1.01,0.4483327865600586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,64,balanced,3.107669194539388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,64,power_law_1.01,0.23479681015014647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,64,power_law_1.01,0.5234943866729737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,32,1,power_law_1.2,4.05491828918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,64,power_law_1.01,0.2532543897628784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,64,power_law_1.2,0.2690752029418945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,64,power_law_1.01,0.28952319622039796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,64,power_law_1.01,0.28424320220947263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,64,balanced,4.09386666615804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,64,power_law_1.01,0.36246399879455565
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,64,balanced,0.06795733173688252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,64,power_law_1.01,0.034835198521614076
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,64,balanced,0.07081066568692525
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,64,balanced,0.07518399755160014
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,64,power_law_1.01,0.03485440015792847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,64,balanced,0.08298666775226593
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,64,balanced,0.09233599901199341
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,64,balanced,0.11236266295115153
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,64,power_law_1.01,0.03495680093765259
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,64,balanced,0.13199999928474426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,64,balanced,0.1728960076967875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,64,power_law_1.01,0.03528960049152374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,32,1,power_law_1.2,5.890361785888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,64,power_law_1.01,0.6984384059906006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,64,power_law_1.01,0.035718399286270144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,64,power_law_1.2,0.2755327939987183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,64,power_law_1.01,0.03775359988212586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,64,power_law_1.01,0.8451392173767089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,64,power_law_1.2,0.2792383909225464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,64,power_law_1.01,0.060761600732803345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,64,power_law_1.01,0.44174718856811523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,64,power_law_1.2,0.2818687915802002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,64,power_law_1.01,1.1614912033081055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,64,power_law_1.2,0.2857919931411743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,64,power_law_1.01,0.5795455932617187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,64,power_law_1.2,0.29212160110473634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,64,balanced,0.2158506711324056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,64,power_law_1.01,0.7095295906066894
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,64,balanced,0.30694399277369183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,64,balanced,0.3898986577987671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,64,power_law_1.01,0.9851327896118164
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,64,balanced,0.5536479949951172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,64,power_law_1.01,0.29190399646759035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,64,balanced,0.7227253119150797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,64,power_law_1.2,0.09200000166893005
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,64,power_law_1.01,0.04597119987010956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,64,power_law_1.01,1.2892288208007812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,64,power_law_1.2,0.08521599769592285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,64,power_law_1.01,0.11519999504089355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,64,power_law_1.2,0.09218559861183166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,64,power_law_1.2,0.08990079760551453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,64,power_law_1.01,0.11798399686813354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,64,power_law_1.2,0.09321600198745728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,64,power_law_1.2,0.09978880286216736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,64,power_law_1.01,0.11831040382385254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,64,power_law_1.01,1.8714752197265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,64,power_law_1.2,0.1382848024368286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,64,power_law_1.2,0.09945600032806397
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,64,power_law_1.01,0.09831039905548096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,64,power_law_1.2,0.11633919477462769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,64,power_law_1.2,0.17111680507659913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,64,power_law_1.01,0.300435209274292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,64,power_law_1.2,0.1545151948928833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,32,1,power_law_1.2,7.716000366210937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,64,power_law_1.01,2.433318328857422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,64,power_law_1.01,0.3168384075164795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,64,power_law_1.01,0.33470079898834226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,64,power_law_1.01,1.509939193725586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,64,power_law_1.2,0.2985343933105469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,64,power_law_1.01,0.3497215986251831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,64,power_law_1.2,0.304915189743042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,64,power_law_1.01,0.39149439334869385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,64,power_law_1.01,3.521779251098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,64,power_law_1.2,0.31861119270324706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,64,power_law_1.01,2.0747007369995116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,64,power_law_1.2,0.33495678901672366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,64,power_law_1.2,0.2584320068359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,64,power_law_1.2,0.3502847909927368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,64,power_law_1.2,0.16859519481658936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,64,power_law_1.2,0.38185598850250246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,64,power_law_1.2,0.16969599723815917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,64,power_law_1.01,2.7249088287353516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,64,power_law_1.2,0.46442880630493166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,64,power_law_1.2,0.17685760259628297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,64,power_law_1.2,0.1784127950668335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,64,power_law_1.2,0.5376448154449462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,64,power_law_1.01,0.08358399868011475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,64,power_law_1.2,0.18751360177993776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,64,power_law_1.01,0.46343679428100587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,64,power_law_1.01,0.0692736029624939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,64,power_law_1.2,0.1963711977005005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,64,power_law_1.01,0.5462207794189453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,64,power_law_1.01,0.06540160179138184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,64,power_law_1.2,0.2014080047607422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,64,power_law_1.01,4.024006271362305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,64,power_law_1.01,0.06501759886741638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,64,power_law_1.2,0.21778559684753418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,64,power_law_1.2,0.2632895946502686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,64,power_law_1.01,0.07013760209083557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,64,power_law_1.2,0.23880319595336913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,64,power_law_1.2,0.2668031930923462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,64,power_law_1.01,0.08763520121574402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,64,power_law_1.2,0.25939838886260985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,64,power_law_1.2,0.27155840396881104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,64,power_law_1.01,0.09847040176391601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,64,power_law_1.01,4.642092895507813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,64,power_law_1.2,0.29592320919036863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,64,balanced,0.04852266609668732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,64,balanced,0.048058668772379555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,64,power_law_1.2,0.37315199375152586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,64,power_law_1.2,0.7463295936584473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,64,power_law_1.2,0.45016961097717284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,64,power_law_1.01,5.235251235961914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,64,balanced,0.045834665497144066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,64,power_law_1.2,0.9123456001281738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,64,balanced,0.044954667488733925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,64,power_law_1.01,0.6961023807525635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,64,balanced,0.04450133442878723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,64,power_law_1.2,0.5946879863739014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,64,balanced,0.09060800075531006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,64,balanced,0.14046933253606161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,64,balanced,0.24116265773773193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,64,power_law_1.01,0.8685952186584472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,64,balanced,0.24445333083470663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,64,power_law_1.2,0.7542975902557373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,64,power_law_1.2,1.2903167724609375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,64,balanced,0.24801599979400635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,64,balanced,0.2515733242034912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,64,balanced,0.2553066611289978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,64,balanced,0.2632533311843872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,64,power_law_1.2,1.0431232452392578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,64,power_law_1.01,0.11649279594421387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,64,balanced,0.04965866605440775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,64,balanced,0.10302399595578511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,64,power_law_1.2,1.6426048278808594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,64,balanced,0.16974933942159018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,64,balanced,0.2393653392791748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,64,balanced,0.2434879938761393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,64,power_law_1.2,1.3723008155822753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,64,balanced,0.24643200635910034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,64,balanced,0.25063467025756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,64,balanced,0.2524320085843404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,64,balanced,0.2616213361422221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,64,power_law_1.2,2.312972831726074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,64,balanced,0.26949866612752277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,64,balanced,0.2757599949836731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,64,power_law_1.2,2.0166400909423827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,64,power_law_1.01,1.1891200065612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,64,power_law_1.2,0.2775424003601074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,64,power_law_1.2,0.03644160032272339
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,64,power_law_1.01,0.140665602684021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,64,power_law_1.2,3.0456640243530275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,64,power_law_1.2,0.03555839955806732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,64,power_law_1.2,2.612063980102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,64,power_law_1.01,1.524550437927246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,64,power_law_1.01,0.1898687958717346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,64,power_law_1.2,0.03644160032272339
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,64,power_law_1.01,0.2522176027297974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,64,power_law_1.2,0.03505919873714447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,64,power_law_1.01,0.3514496088027954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,64,power_law_1.2,0.03560320138931274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,64,balanced,0.28964267174402875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,64,power_law_1.01,2.1886079788208006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,64,balanced,0.30390934149424237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,64,balanced,0.318229337533315
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,64,power_law_1.01,0.4521984100341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,64,balanced,0.3473973274230957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,64,balanced,0.27086400985717773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,64,power_law_1.2,3.9334976196289064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,64,power_law_1.2,4.658310317993164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,64,balanced,0.4039253393809001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,64,balanced,0.27929067611694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,64,balanced,0.29264533519744873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,64,power_law_1.2,0.28469760417938234
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,64,power_law_1.01,0.6834496021270752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,64,balanced,0.306058665116628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,64,balanced,0.3219040036201477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,64,power_law_1.2,0.29374721050262453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,64,power_law_1.01,2.850739288330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,64,power_law_1.01,0.8637056350708008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,64,power_law_1.2,0.3010432004928589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,64,balanced,0.06684799989064534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,64,balanced,0.06942399839560191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,64,power_law_1.2,0.3164799928665161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,64,balanced,0.07330133517583211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,64,balanced,0.07391466697057088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,64,balanced,0.09614400068918864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,64,power_law_1.2,0.33777918815612795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,64,balanced,0.1530933380126953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,64,power_law_1.2,0.03989120125770569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,64,balanced,0.15806933244069418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,64,balanced,0.1592639982700348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,64,power_law_1.2,0.3566335916519165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,64,balanced,0.1644213298956553
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,64,power_law_1.2,0.05872640013694763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,64,power_law_1.2,6.123635101318359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,64,balanced,0.46107200781504315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,64,power_law_1.2,0.39548161029815676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,64,power_law_1.2,0.04859519898891449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,64,balanced,0.574181318283081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,64,balanced,0.35227731863657635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,64,power_law_1.2,0.4738111972808838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,64,balanced,0.6860106786092123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,64,balanced,0.41467734177907306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,64,balanced,0.47044801712036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,64,balanced,0.9225653012593588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,64,power_law_1.2,0.5655744075775146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,64,balanced,1.143818696339925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,64,power_law_1.2,0.765177583694458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,64,balanced,1.6306773821512859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,64,balanced,0.16666133205095926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,64,balanced,0.17485332489013672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,64,power_law_1.2,5.402822494506836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,64,balanced,0.18312533696492514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,64,balanced,0.18930133183797201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,64,power_law_1.2,0.9364928245544434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,64,power_law_1.01,0.09523199796676636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,64,power_law_1.2,0.1219648003578186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,64,power_law_1.01,0.09568639993667602
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,64,power_law_1.2,0.1161728024482727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,64,power_law_1.2,1.3024191856384277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,64,balanced,0.5875946680704752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,64,power_law_1.2,0.11015039682388306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,64,balanced,0.7054346402486166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,64,power_law_1.2,0.09327359795570374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,64,balanced,0.949461301167806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,64,power_law_1.2,0.0896511971950531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,64,power_law_1.2,1.6708608627319337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,64,power_law_1.01,4.165856170654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,64,balanced,1.1841440200805664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,64,balanced,0.20137600104014078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,64,balanced,2.1247146924336753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,64,balanced,1.677984078725179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,64,power_law_1.2,2.5201215744018555
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,64,balanced,0.029311999678611755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,64,power_law_1.01,0.095769602060318
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,64,balanced,0.029653333127498627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,64,balanced,0.030080000559488933
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,64,balanced,0.0305226668715477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,64,balanced,2.1752427419026694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,64,balanced,3.1338561375935874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,64,power_law_1.2,0.06873599886894226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,64,power_law_1.01,5.66302719116211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,64,power_law_1.2,3.1358591079711915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,64,balanced,3.157594680786133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,64,power_law_1.2,0.06595199704170226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,64,power_law_1.01,0.08386560082435608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,64,balanced,4.148389180501302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,64,power_law_1.01,0.08288000226020813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,64,power_law_1.01,0.07859200239181519
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,64,balanced,0.03359466542800268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,64,balanced,4.145354588826497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,64,power_law_1.2,5.048281478881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,64,balanced,0.2179786761601766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,64,balanced,0.23201066255569458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,64,power_law_1.2,0.06730239987373351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,64,balanced,0.2615999976793925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,64,balanced,0.3187306722005208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,64,power_law_1.2,0.07218559980392455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,64,power_law_1.01,0.08433279991149903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,64,power_law_1.2,0.08670719861984252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,64,power_law_1.01,0.14632320404052734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,64,power_law_1.01,0.08376960158348083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,64,power_law_1.01,0.12025599479675293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,64,power_law_1.2,0.09734399914741516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,64,power_law_1.01,0.17297279834747314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,64,power_law_1.01,0.09127039909362793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,64,power_law_1.01,0.1504320025444031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,64,power_law_1.2,0.12072319984436035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,64,power_law_1.01,0.2362623929977417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,64,power_law_1.01,0.09080320000648498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,64,power_law_1.01,0.23749759197235107
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,64,power_law_1.2,0.14769279956817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,64,power_law_1.01,0.2424704074859619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,64,power_law_1.2,6.684243011474609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,64,power_law_1.01,0.24298880100250245
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,64,power_law_1.2,0.2055232048034668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,64,power_law_1.01,0.24365439414978027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,64,power_law_1.2,0.09488639831542969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,64,power_law_1.01,0.24500479698181152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,64,power_law_1.01,0.24805119037628173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,64,power_law_1.2,0.09428480267524719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,64,balanced,0.37489600976308185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,64,power_law_1.01,0.24839038848876954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,64,power_law_1.01,0.25157120227813723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,64,power_law_1.2,0.09530879855155945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,64,balanced,0.04176533222198486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,64,balanced,0.04783466458320618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,64,power_law_1.01,0.2597055912017822
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,64,balanced,0.06285333136717479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,64,power_law_1.2,0.14422399997711183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,64,balanced,0.1172213355700175
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,64,balanced,0.11754666765530904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,64,power_law_1.01,0.2672703981399536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,64,balanced,0.11126400033632915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,64,balanced,0.09551466504732768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,64,power_law_1.01,0.2745984077453613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,64,power_law_1.01,0.10583679676055908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,64,power_law_1.01,0.28688640594482423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,64,power_law_1.01,0.03285120129585266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,64,power_law_1.01,0.14867199659347535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,64,power_law_1.01,0.3030783891677856
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,64,power_law_1.2,0.26873600482940674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,64,power_law_1.01,0.03235200047492981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,64,power_law_1.01,0.15491199493408203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,64,power_law_1.01,0.31845760345458984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,64,balanced,0.4888746738433838
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,64,power_law_1.01,0.03216640055179596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,64,power_law_1.01,0.1572160005569458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,64,power_law_1.01,0.3504767894744873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,64,power_law_1.01,0.25242879390716555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,64,balanced,0.6004693508148193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,64,power_law_1.2,0.17311999797821045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,64,power_law_1.01,0.2606271982192993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,64,power_law_1.01,0.4120639801025391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,64,power_law_1.01,0.1617535948753357
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,64,balanced,0.08785600463549297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,64,balanced,0.844042698542277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,64,power_law_1.01,0.26860799789428713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,64,balanced,0.07042133311430614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,64,power_law_1.2,0.23573119640350343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,64,power_law_1.01,0.1634112000465393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,64,power_law_1.01,0.4818943977355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,64,balanced,0.07039999961853027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,64,power_law_1.01,0.2756736040115356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,64,balanced,0.0633493314186732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,64,balanced,1.0667146841684978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,64,power_law_1.01,0.17118719816207886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,64,power_law_1.2,0.24147839546203614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,64,balanced,0.06892266869544983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,64,power_law_1.01,0.6449151992797851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,64,power_law_1.01,0.2905791997909546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,64,power_law_1.01,0.17941759824752807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,64,power_law_1.2,0.2433919906616211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,64,balanced,1.526528040568034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,64,power_law_1.2,0.3783423900604248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,64,power_law_1.01,0.18468480110168456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,64,power_law_1.2,0.2479423999786377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,64,power_law_1.01,0.7829760074615478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,64,power_law_1.01,0.2012160062789917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,64,power_law_1.2,0.24987521171569824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,64,power_law_1.2,0.5123263835906983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,64,balanced,1.9870346387227376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,64,power_law_1.01,0.21424639225006104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,64,power_law_1.2,0.2585472106933594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,64,power_law_1.01,0.03242239952087402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,64,power_law_1.01,0.23228800296783447
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,64,power_law_1.2,0.7139328002929688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,64,power_law_1.2,0.26776320934295655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,64,power_law_1.01,0.03195520043373108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,64,power_law_1.01,0.2674623966217041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,64,power_law_1.2,0.2738624095916748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,64,balanced,2.910223960876465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,64,power_law_1.01,0.036134400963783266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,64,power_law_1.01,0.3329024076461792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,64,power_law_1.2,0.2884416103363037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,64,power_law_1.2,1.07073917388916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,64,power_law_1.2,0.3039423942565918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,64,power_law_1.01,0.40556159019470217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,64,power_law_1.2,0.0815999984741211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,64,power_law_1.01,0.047660800814628604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,64,power_law_1.2,0.3198784112930298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,64,power_law_1.2,0.08311039805412293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,64,power_law_1.01,0.533184003829956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,64,balanced,3.838015874226888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,64,power_law_1.2,0.34997758865356443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,64,power_law_1.01,1.0712832450866698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,64,power_law_1.2,0.08303999900817871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,64,power_law_1.01,0.6555647850036621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,64,power_law_1.2,0.41930241584777833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,64,power_law_1.2,0.12058240175247192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,64,power_law_1.01,1.3742079734802246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,64,power_law_1.2,0.14816639423370362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,64,power_law_1.01,0.9251520156860351
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,64,balanced,0.0763626645008723
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,64,balanced,0.08481066425641377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,64,power_law_1.2,0.23765759468078612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,64,balanced,0.1032480001449585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,64,power_law_1.01,0.3043519973754883
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,64,balanced,0.11954666177431743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,64,power_law_1.01,1.1887807846069336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,64,balanced,0.15428800384203592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,64,power_law_1.01,1.9769472122192382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,64,balanced,0.19207467635472616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,64,balanced,0.2731253306070964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,64,balanced,0.3464266856511434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,64,power_law_1.01,1.7061311721801757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,64,power_law_1.2,0.07800319790840149
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,64,balanced,0.48955734570821124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,64,power_law_1.01,2.502592086791992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,64,power_law_1.2,0.08177919983863831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,64,balanced,0.6365973154703776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,64,power_law_1.2,0.5011136054992675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,64,power_law_1.2,0.09085440039634704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,64,power_law_1.2,0.685484790802002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,64,power_law_1.01,2.2252607345581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,64,power_law_1.2,0.2419327974319458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,64,power_law_1.01,0.3200704097747803
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,64,power_law_1.01,0.052723199129104614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,64,power_law_1.2,0.2458751916885376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,64,power_law_1.01,3.756422424316406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,64,power_law_1.01,0.35459840297698975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,64,power_law_1.01,0.10906879901885987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,64,power_law_1.01,0.421068811416626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,64,power_law_1.01,3.2708736419677735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,64,power_law_1.01,0.10532480478286743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,64,power_law_1.01,0.4879551887512207
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,64,power_law_1.01,0.09574400186538697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,64,power_law_1.01,0.0895359992980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,64,power_law_1.2,0.08885759711265565
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,64,power_law_1.01,0.08098559975624084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,64,power_law_1.2,0.10348800420761109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,64,power_law_1.2,0.8277952194213867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,64,power_law_1.01,0.06287360191345215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,64,power_law_1.01,4.894790267944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,64,power_law_1.2,0.1412608027458191
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,64,power_law_1.01,0.0654528021812439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,64,power_law_1.01,4.3635009765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,64,power_law_1.2,0.24929919242858886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,64,power_law_1.2,0.1555840015411377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,64,power_law_1.01,0.05935360193252563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,64,power_law_1.2,0.2530751943588257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,64,power_law_1.2,0.15433599948883056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,64,power_law_1.2,0.2612096071243286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,64,power_law_1.2,0.16183040142059327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,64,power_law_1.2,0.269868803024292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,64,power_law_1.01,0.643116807937622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,64,power_law_1.2,0.16409599781036377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,64,power_law_1.2,0.2767616033554077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,64,power_law_1.2,0.17160320281982422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,64,power_law_1.01,0.7813119888305664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,64,power_law_1.2,0.2916352033615112
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,64,power_law_1.2,0.03330560028553009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,64,power_law_1.2,0.1798848032951355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,64,balanced,0.027327999472618103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,64,balanced,0.027589333554108936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,64,power_law_1.2,0.03278079926967621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,64,power_law_1.2,0.18652160167694093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,64,power_law_1.01,1.085689640045166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,64,power_law_1.2,1.1070207595825194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,64,power_law_1.2,0.19893759489059448
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,64,power_law_1.01,0.06058239936828613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,64,power_law_1.2,0.22163839340209962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,64,power_law_1.01,1.363046360015869
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,64,power_law_1.01,0.07767680287361145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,64,power_law_1.2,0.23796479701995848
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,64,power_law_1.01,0.08721280097961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,64,power_law_1.2,0.27270400524139404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,64,power_law_1.01,0.10844800472259522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,64,power_law_1.2,0.350816011428833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,64,power_law_1.01,1.9845632553100585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,64,power_law_1.01,0.12732160091400146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,64,power_law_1.2,0.41564159393310546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,64,power_law_1.2,0.3090559959411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,64,balanced,0.039066667358080544
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,64,power_law_1.01,0.16981760263442994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,64,balanced,0.029557332396507263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,64,power_law_1.2,0.3242815971374512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,64,power_law_1.2,0.5495039939880371
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,64,power_law_1.01,0.21911680698394775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,64,power_law_1.2,0.3615936040878296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,64,power_law_1.01,2.576153564453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,64,power_law_1.01,0.310534405708313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,64,power_law_1.2,0.6818431854248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,64,power_law_1.2,1.4550463676452636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,64,power_law_1.2,0.43351039886474607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,64,power_law_1.01,0.4017792224884033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,64,power_law_1.2,0.9787136077880859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,64,power_law_1.2,0.5139711856842041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,64,power_law_1.01,0.5916607856750489
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,64,power_law_1.2,2.0232576370239257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,64,power_law_1.2,0.690451192855835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,64,power_law_1.2,1.2286335945129394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,64,power_law_1.01,3.704595184326172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,64,balanced,0.032602667808532715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,64,power_law_1.01,0.7755199909210205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,64,balanced,0.039706667264302574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,64,balanced,0.04389866689840952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,64,balanced,0.03962666789690653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,64,balanced,0.06025599936644236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,64,balanced,0.04072533299525579
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,64,balanced,0.06001600126425425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,64,balanced,0.042090664307276406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,64,balanced,0.062277331948280334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,64,balanced,0.0421013335386912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,64,balanced,0.06233066817124685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,64,balanced,0.04528533418973287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,64,balanced,0.06598400076230367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,64,power_law_1.2,2.916697692871094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,64,balanced,0.06682666639486949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,64,balanced,0.06859733164310455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,64,balanced,0.07260799904664357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,64,balanced,0.07635200023651123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,64,balanced,0.0828000009059906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,64,balanced,0.08895466725031535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,64,balanced,0.09436800082524617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,64,balanced,0.10735999544461568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,64,balanced,0.12963733077049255
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,64,power_law_1.01,4.972998428344726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,64,balanced,0.042768001556396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,64,balanced,0.0563679983218511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,64,balanced,0.05608533322811127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,64,power_law_1.2,0.03287039995193482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,64,balanced,0.060133333007494606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,64,balanced,0.060165335734685264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,64,power_law_1.2,4.288371276855469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,64,power_law_1.2,0.03287039995193482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,64,power_law_1.2,1.8003583908081056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,64,balanced,0.04696000119050344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,64,power_law_1.2,0.03399679958820343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,64,balanced,0.05091199775536855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,64,balanced,0.04958933095137278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,64,balanced,0.05346666773160299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,64,balanced,0.05249066650867462
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,64,power_law_1.2,0.038438400626182555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,64,balanced,0.05690133571624756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,64,balanced,0.058517331878344216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,64,balanced,0.06132799883683523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,64,power_law_1.2,0.042310398817062375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,64,balanced,0.06670400003592174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,64,balanced,0.07451733450094859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,64,balanced,0.1527359982331594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,64,balanced,0.07909866670767467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,64,power_law_1.2,0.05501440167427063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,64,power_law_1.2,0.8281984329223633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,64,power_law_1.2,0.10101759433746338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,64,balanced,0.06218666831652323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,64,power_law_1.2,5.594188690185547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,64,balanced,0.06431999802589417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,64,power_law_1.2,0.10819840431213379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,64,power_law_1.01,0.026547199487686156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,64,power_law_1.2,1.1701120376586913
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,64,power_law_1.2,0.08606719970703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,64,power_law_1.01,0.02643199861049652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,64,power_law_1.2,0.08446720242500305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,64,power_law_1.01,0.03824000060558319
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,64,power_law_1.2,0.07440639734268188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,64,power_law_1.2,1.4857024192810058
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,64,balanced,0.025546667476495106
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,64,power_law_1.2,0.0635968029499054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,64,power_law_1.2,2.408032035827637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,64,balanced,0.1994933287302653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,64,power_law_1.2,0.06538239717483521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,64,balanced,0.23677867650985718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,64,balanced,0.33350932598114014
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,64,power_law_1.2,0.058745598793029784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,64,power_law_1.2,2.2255168914794923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,64,power_law_1.2,0.0685375988483429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,64,balanced,0.0682666649421056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,64,balanced,0.07030933101971944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,64,balanced,0.072543998559316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,64,power_law_1.2,0.0821120023727417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,64,power_law_1.2,0.09160959720611572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,64,power_law_1.2,3.5299774169921876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,64,power_law_1.2,0.11248639822006226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,64,power_law_1.01,0.04208639860153198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,64,power_law_1.2,2.908537673950195
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,64,balanced,0.02552533398071925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,64,power_law_1.01,0.0543615996837616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,64,balanced,0.025306666890780132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,64,balanced,0.09085333347320557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,64,power_law_1.01,0.05539199709892273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,64,power_law_1.01,0.058387202024459836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,64,power_law_1.01,0.05836799740791321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,64,balanced,0.42240532239278156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,64,power_law_1.01,0.06090239882469177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,64,balanced,0.5946773290634155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,64,balanced,0.07891199986139934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,64,power_law_1.2,4.709036636352539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,64,balanced,0.7699466546376547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,64,power_law_1.2,4.174655914306641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,64,balanced,1.1167199611663818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,64,power_law_1.01,0.029203200340270997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,64,power_law_1.01,0.028403198719024657
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,64,power_law_1.2,0.1355839967727661
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,64,balanced,0.025386666258176167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,64,power_law_1.01,0.03934719860553741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,64,balanced,0.11441066861152649
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,64,balanced,0.02754666656255722
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,64,balanced,0.02962133288383484
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,64,power_law_1.2,0.1811519980430603
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,64,balanced,0.0356480007370313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,64,balanced,0.03738666574160258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,64,balanced,0.03766933331886927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,64,power_law_1.2,0.24078719615936278
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,64,power_law_1.01,0.06220800280570984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,64,balanced,0.08514133095741272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,64,balanced,0.09296000003814697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,64,balanced,0.10496000448862712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,64,balanced,0.12689600388209024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,64,power_law_1.2,5.642752075195313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,64,balanced,0.14882133404413858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,64,balanced,0.19010132551193237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,64,balanced,0.22317866484324136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,64,power_law_1.01,0.03684479892253876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,64,balanced,0.3123679955800374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,64,balanced,1.4706986745198567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,64,balanced,0.39479998747507733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,64,power_law_1.01,0.03918719887733459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,64,balanced,0.5537013212839762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,64,power_law_1.01,0.04433279931545257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,64,power_law_1.01,0.04135040044784546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,64,balanced,0.7118720213572184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,64,power_law_1.01,0.04058879911899567
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,64,balanced,0.03766400118668874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,64,balanced,0.038015998899936676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,64,balanced,0.03995733211437861
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,64,balanced,0.040394666294256844
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,64,power_law_1.2,0.33048319816589355
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,64,balanced,0.03762666632731756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,64,power_law_1.01,0.06535040140151978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,64,power_law_1.01,0.06872959733009339
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,64,power_law_1.2,0.4565311908721924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,64,power_law_1.01,0.07151359915733338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,64,power_law_1.01,0.02343679964542389
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,64,power_law_1.2,0.6603007793426514
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,64,power_law_1.01,0.023027199506759643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,64,balanced,0.13983999689420065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,64,power_law_1.01,0.022969600558280946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,64,power_law_1.01,0.05689600110054016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,64,power_law_1.2,0.8311103820800781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,64,power_law_1.01,0.0236735999584198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,64,power_law_1.01,0.05811200141906738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,64,balanced,1.034010648727417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,64,power_law_1.01,0.025497600436210632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,64,power_law_1.01,0.06093440055847168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,64,balanced,0.03755733370780945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,64,power_law_1.01,0.02717440128326416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,64,balanced,0.037989333271980286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,64,balanced,1.3544480005900066
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,64,balanced,0.038346665600935616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,64,balanced,0.0418453315893809
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,64,power_law_1.01,0.03407360017299652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,64,balanced,0.043866669138272606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,64,power_law_1.2,0.0308351993560791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,64,balanced,0.04866133133570353
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,64,power_law_1.01,0.034796801209449765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,64,balanced,0.05412266651789347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,64,balanced,0.06433600187301636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,64,power_law_1.01,0.07760639786720276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,64,balanced,0.07502933343251546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,64,balanced,0.10131200154622395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,64,balanced,0.13590400417645773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,64,balanced,0.19029333194096884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,64,balanced,0.2499199906984965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,64,power_law_1.01,0.04586879909038544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,64,power_law_1.2,0.03752320110797882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,64,power_law_1.2,0.026086398959159852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,64,power_law_1.01,0.04581120014190674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,64,power_law_1.2,0.03849599957466125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,64,power_law_1.01,0.06188160181045532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,64,power_law_1.01,0.04663679897785187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,64,power_law_1.2,0.041043201088905336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,64,power_law_1.01,0.04806399941444397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,64,power_law_1.2,0.041126400232315063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,64,power_law_1.01,0.04979200065135956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,64,power_law_1.2,0.028870400786399842
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,64,power_law_1.01,0.035206401348114015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,64,power_law_1.2,0.046188798546791074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,64,power_law_1.01,0.05134720206260681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,64,power_law_1.2,0.03948799967765808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,64,power_law_1.01,0.08375040292739869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,64,power_law_1.2,0.04528000056743622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,64,power_law_1.01,0.05421440005302429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,64,power_law_1.2,0.04477440118789673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,64,balanced,0.17323732376098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,64,power_law_1.01,0.058182400465011594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,64,balanced,0.21444799502690634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,64,power_law_1.2,0.057843202352523805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,64,power_law_1.01,0.05845119953155518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,64,power_law_1.2,0.058924800157546996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,64,power_law_1.01,0.0657472014427185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,64,power_law_1.2,0.06165120005607605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,64,power_law_1.01,0.07203840017318726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,64,power_law_1.2,0.06236159801483154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,64,power_law_1.01,0.06304640173912049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,64,power_law_1.2,0.06364799737930298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,64,power_law_1.01,0.0648959994316101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,64,power_law_1.2,0.06539520025253295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,64,power_law_1.01,0.06857600212097167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,64,power_law_1.2,0.06780160069465638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,64,power_law_1.01,0.035667198896408084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,64,power_law_1.01,0.07184000015258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,64,power_law_1.2,0.07228800058364868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,64,power_law_1.01,0.0907904028892517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,64,power_law_1.01,0.07331839799880982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,64,power_law_1.2,0.0754047989845276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,64,power_law_1.01,0.10206079483032227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,64,power_law_1.01,0.08072959780693054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,64,balanced,0.30189333359400433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,64,power_law_1.2,0.07987200021743775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,64,power_law_1.01,0.1280832052230835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,64,power_law_1.01,0.0871999979019165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,64,balanced,0.38576531410217285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,64,power_law_1.2,0.08559359908103943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,64,power_law_1.01,0.09266560077667237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,64,balanced,0.5454453229904175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,64,power_law_1.2,0.026771199703216553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,64,power_law_1.2,0.09164159893989562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,64,power_law_1.01,0.10514559745788574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,64,balanced,0.7069599628448486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,64,power_law_1.2,0.037651199102401736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,64,power_law_1.01,0.12564480304718018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,64,power_law_1.2,0.042028799653053284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,64,balanced,1.0288693110148113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,64,power_law_1.01,0.15268479585647582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,64,power_law_1.2,0.05411199927330017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,64,power_law_1.01,0.19857920408248902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,64,power_law_1.2,0.05539839863777161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,64,balanced,1.3493173917134602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,64,power_law_1.2,0.048281601071357726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,64,power_law_1.2,0.05788800120353699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,64,power_law_1.2,0.058508801460266116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,64,power_law_1.2,0.06082559823989868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,64,power_law_1.2,0.062080001831054686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,64,power_law_1.01,0.1471935987472534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,64,power_law_1.2,0.06492159962654113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,64,power_law_1.01,0.18731520175933838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,64,power_law_1.01,0.0794431984424591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,64,power_law_1.2,0.10219520330429077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,64,power_law_1.2,0.0686784029006958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,64,power_law_1.01,0.22504959106445313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,64,power_law_1.01,0.0896511971950531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,64,power_law_1.2,0.12460160255432129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,64,power_law_1.2,0.07095040082931518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,64,power_law_1.01,0.11346559524536133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,64,power_law_1.2,0.077292799949646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,64,power_law_1.2,0.08407040238380432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,64,power_law_1.2,0.022598400712013245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,64,power_law_1.01,0.24372479915618897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,64,power_law_1.2,0.08981119990348815
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,64,power_law_1.2,0.022944000363349915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,64,power_law_1.2,0.10204160213470459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,64,power_law_1.01,0.34853758811950686
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,64,power_law_1.2,0.023039999604225158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,64,power_law_1.2,0.12458879947662353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,64,power_law_1.01,0.45090560913085936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,64,power_law_1.2,0.02338559925556183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,64,power_law_1.2,0.024588799476623534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,64,power_law_1.01,0.6492224216461182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,64,power_law_1.2,0.02743679881095886
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,64,power_law_1.2,0.0336896002292633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,64,power_law_1.01,0.32159359455108644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,64,power_law_1.01,0.826534366607666
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,64,power_law_1.2,0.03492479920387268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,64,power_law_1.01,0.14085760116577148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,64,power_law_1.01,0.42081918716430666
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,64,power_law_1.2,0.035257598757743834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,64,power_law_1.01,0.17806719541549682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,64,power_law_1.01,1.2385343551635741
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,64,power_law_1.2,0.0353408008813858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,64,power_law_1.01,0.5976640224456787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,64,power_law_1.01,0.22396159172058105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,64,power_law_1.2,0.03620480000972748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,64,power_law_1.01,0.3155904054641724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,64,power_law_1.2,0.14926719665527344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,64,power_law_1.01,0.7857024192810058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,64,power_law_1.01,1.6260223388671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,64,power_law_1.01,0.4044928073883057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,64,balanced,0.035599999129772186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,64,balanced,0.033904001116752625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,64,balanced,0.05825066566467285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,64,balanced,0.06639466683069865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,64,power_law_1.01,0.5931647777557373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,64,balanced,0.10014399886131287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,64,power_law_1.2,0.14970879554748534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,64,power_law_1.01,0.7495423793792725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,64,power_law_1.2,0.20452480316162108
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,64,power_law_1.01,0.0359360009431839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,64,power_law_1.2,0.2479680061340332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,64,power_law_1.01,0.036524799466133115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,64,power_law_1.01,1.1020288467407227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,64,power_law_1.01,0.03624959886074066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,64,power_law_1.2,0.3630208015441895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,64,power_law_1.2,0.036550399661064145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,64,power_law_1.01,0.03600000143051148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,64,power_law_1.2,0.1877568006515503
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,64,power_law_1.2,0.03713279962539673
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,64,power_law_1.01,0.03703039884567261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,64,power_law_1.01,1.4892224311828612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,64,power_law_1.01,0.03699199855327606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,64,balanced,0.03569599986076355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,64,balanced,0.035786665976047516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,64,power_law_1.01,0.03768320083618164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,64,power_law_1.2,0.23690879344940186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,64,power_law_1.01,0.04120959937572479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,64,power_law_1.2,0.3314239978790283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,64,power_law_1.01,0.042828801274299624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,64,power_law_1.2,0.048563200235366824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,64,power_law_1.01,0.04837760031223297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,64,power_law_1.2,0.050835198163986205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,64,power_law_1.2,0.43633279800415037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,64,power_law_1.2,0.03649280071258545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,64,power_law_1.01,1.1377087593078614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,64,power_law_1.2,0.47226881980895996
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,64,power_law_1.2,0.03711360096931458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,64,power_law_1.2,0.03811199963092804
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,64,power_law_1.2,0.038227200508117676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,64,power_law_1.01,1.5117440223693848
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,64,power_law_1.2,0.041715198755264284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,64,balanced,0.049642667174339294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,64,balanced,0.05644799768924713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,64,power_law_1.2,0.04405759871006012
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,64,power_law_1.2,0.05145599842071533
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,64,power_law_1.01,0.05443199872970581
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,64,power_law_1.2,0.0586624026298523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,64,power_law_1.01,0.06855040192604064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,64,power_law_1.2,0.05088000297546387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,64,balanced,0.058287998040517174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,64,balanced,0.06821866830190022
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,64,power_law_1.2,0.07578880190849305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,64,power_law_1.2,0.05428479909896851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,64,balanced,0.09718933701515198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,64,power_law_1.2,0.05740159749984741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,64,balanced,0.09909333785374959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,64,balanced,0.10126399993896484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,64,balanced,0.10534399747848511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,64,balanced,0.10739733775456746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,64,balanced,0.10931199789047241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,64,balanced,0.11341333389282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,64,balanced,0.11967999736467998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,64,power_law_1.2,0.09534080028533935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,64,balanced,0.12777066230773926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,64,balanced,0.1381653348604838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,64,balanced,0.14834133783976236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,64,balanced,0.10640000303586324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,64,balanced,0.10560533404350281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,64,balanced,0.11486400167147319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,64,balanced,0.11355200409889221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,64,balanced,0.11611732840538025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,64,balanced,0.1300373375415802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,64,balanced,0.13492266337076822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,64,balanced,0.1410719950993856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,64,balanced,0.1525759994983673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,64,power_law_1.2,0.05989760160446167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,64,balanced,0.16110400358835855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,64,balanced,0.17149333159128824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,64,power_law_1.2,0.0657472014427185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,64,power_law_1.2,0.07360640168190002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,64,power_law_1.2,0.6299007892608642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,64,power_law_1.2,0.07959679961204529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,64,power_law_1.2,0.09054080247879029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,64,balanced,0.16082132856051126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,64,power_law_1.2,0.11738879680633545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,64,balanced,0.18674665689468384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,64,balanced,0.23082133134206137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,64,power_law_1.01,0.0847487986087799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,64,balanced,0.27985600630442303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,64,power_law_1.2,0.6862400054931641
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,64,power_law_1.01,0.1221824049949646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,64,balanced,0.05727999905745188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,64,balanced,0.05921066800753275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,64,balanced,0.06925866504510243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,64,balanced,0.19125332434972128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,64,balanced,0.22791999578475952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,64,balanced,0.2701546748479207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,64,balanced,0.3635786771774292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,64,balanced,0.44998399416605633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,64,balanced,0.6208959817886353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,64,power_law_1.2,0.14177919626235963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,64,power_law_1.2,0.18649599552154542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,64,power_law_1.2,0.2315648078918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,64,balanced,0.07030933101971944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,64,balanced,0.07423999905586243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,64,balanced,0.0753600001335144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,64,power_law_1.2,0.32953600883483886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,64,balanced,0.7904106775919596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,64,balanced,0.37854933738708496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,64,balanced,1.1349120140075684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,64,balanced,0.4710186719894409
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,64,power_law_1.01,0.15660799741744996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,64,balanced,0.6527946790059408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,64,power_law_1.2,0.4239488124847412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,64,balanced,0.8339893023173014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,64,balanced,0.07829333345095317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,64,balanced,0.07909333209196727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,64,power_law_1.2,0.609548807144165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,64,balanced,0.0844533344109853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,64,balanced,1.2021386623382568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,64,balanced,1.5741653442382812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,64,balanced,2.3144853909810386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,64,balanced,1.4810506502787273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,64,balanced,3.0546401341756186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,64,balanced,2.196394602457682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,64,balanced,2.8977654774983725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,64,power_law_1.01,0.03878400027751923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,64,power_law_1.01,0.21521921157836915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,64,power_law_1.01,0.039417600631713866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,64,power_law_1.01,0.2870975971221924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,64,power_law_1.2,0.8748288154602051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,64,balanced,0.09124267101287842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,64,balanced,0.09635733564694722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,64,power_law_1.01,0.07781760096549988
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,64,balanced,0.10560533404350281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,64,balanced,0.11682666341463725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,64,balanced,0.12727466225624084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,64,power_law_1.01,0.07724159955978394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,64,balanced,0.15018666783968607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,64,balanced,0.19279466072718301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,64,power_law_1.01,0.10076800584793091
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,64,balanced,0.23900800943374634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,64,balanced,0.3356586694717407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,64,power_law_1.01,0.10695680379867553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,64,balanced,0.42130664984385174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,64,power_law_1.01,0.1069375991821289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,64,balanced,0.5964426596959432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,64,power_law_1.01,0.11040639877319336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,64,power_law_1.01,0.1173248052597046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,64,power_law_1.01,0.1193343997001648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,64,balanced,0.025434667865435284
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,64,balanced,0.025477332373460133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,64,power_law_1.01,0.12662400007247926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,64,balanced,0.025600001215934753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,64,balanced,0.025402667621771496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,64,balanced,0.02956799914439519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,64,power_law_1.01,0.13308800458908082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,64,balanced,0.03164266546567281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,64,balanced,0.03978666663169861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,64,power_law_1.01,0.1410367965698242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,64,balanced,0.04188266893227895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,64,balanced,0.058389330903689064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,64,power_law_1.2,0.1334272027015686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,64,power_law_1.01,0.1511615991592407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,64,power_law_1.01,0.16147840023040771
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,64,power_law_1.2,0.16839040517807008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,64,power_law_1.01,0.17073919773101806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,64,power_law_1.01,0.1914240002632141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,64,power_law_1.01,0.03781760036945343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,64,power_law_1.01,0.23022079467773438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,64,power_law_1.01,0.04099839925765991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,64,power_law_1.01,0.2796927928924561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,64,power_law_1.01,0.05864959955215454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,64,power_law_1.01,0.06828160285949707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,64,power_law_1.01,0.39559040069580076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,64,power_law_1.01,0.09545599818229675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,64,power_law_1.01,0.4921408176422119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,64,power_law_1.01,0.09690240025520325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,64,power_law_1.01,0.09995520114898682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,64,power_law_1.01,0.7111936092376709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,64,balanced,0.7684160073598226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,64,power_law_1.01,0.10227199792861938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,64,power_law_1.01,0.10481280088424683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,64,balanced,1.1097013155619304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,64,power_law_1.01,0.9341695785522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,64,power_law_1.01,0.10728960037231446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,64,power_law_1.01,0.11265920400619507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,64,balanced,1.4538720448811848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,64,power_law_1.01,0.1182528018951416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,64,power_law_1.01,1.365401554107666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,64,power_law_1.01,0.1245695948600769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,64,power_law_1.01,0.13593599796295167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,64,balanced,2.1336053212483725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,64,power_law_1.01,0.1481727957725525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,64,power_law_1.01,1.7823936462402343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,64,power_law_1.01,0.15973119735717772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,64,power_law_1.01,0.18334720134735108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,64,balanced,2.818863868713379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,64,power_law_1.01,0.22962560653686523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,64,power_law_1.2,0.7993152141571045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,64,power_law_1.01,0.05431039929389954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,64,power_law_1.01,0.2828160047531128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,64,power_law_1.2,0.8350655555725097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,64,power_law_1.01,0.055667197704315184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,64,power_law_1.01,2.689740753173828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,64,power_law_1.01,0.40032639503479006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,64,power_law_1.01,0.05928959846496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,64,power_law_1.01,0.05958399772644043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,64,power_law_1.01,0.06849279999732971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,64,power_law_1.01,0.0707264006137848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,64,power_law_1.01,0.07184640169143677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,64,power_law_1.01,3.614790344238281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,64,power_law_1.01,0.07549440264701843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,64,power_law_1.01,0.07690240144729614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,64,power_law_1.01,0.07870720028877258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,64,power_law_1.01,0.08412160277366638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,64,power_law_1.01,0.08918399810791015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,64,power_law_1.01,0.5049024105072022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,64,power_law_1.01,0.09704959988594056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,64,power_law_1.01,0.10925439596176148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,64,power_law_1.01,0.7380159854888916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,64,power_law_1.01,0.12052479982376099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,64,power_law_1.01,0.1311360001564026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,64,power_law_1.01,0.9144127845764161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,64,power_law_1.01,0.16057599782943727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,64,power_law_1.01,0.20746879577636718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,64,power_law_1.01,1.3809151649475098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,64,power_law_1.01,0.2624255895614624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,64,balanced,0.05819733440876007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,64,power_law_1.01,0.36840319633483887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,64,balanced,0.04195733368396759
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,64,balanced,0.04142933338880539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,64,balanced,0.041877334316571556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,64,balanced,0.041802664597829185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,64,power_law_1.01,1.82293758392334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,64,balanced,0.04418666660785675
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,64,balanced,0.04498666524887085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,64,balanced,0.04584533472855886
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,64,balanced,0.05009600023428599
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,64,power_law_1.01,0.024460799992084503
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,64,balanced,0.05418133238951365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,64,balanced,0.06434666613737743
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,64,balanced,0.07460266848405202
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,64,power_law_1.01,0.02447360008955002
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,64,balanced,0.10426132877667744
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,64,balanced,0.13397866487503052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,64,power_law_1.01,0.024985599517822265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,64,balanced,0.18862400452295938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,64,power_law_1.01,2.7948863983154295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,64,power_law_1.01,0.02529279887676239
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,64,power_law_1.01,0.02800000011920929
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,64,power_law_1.01,0.030803200602531434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,64,power_law_1.01,0.46921601295471194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,64,power_law_1.01,0.03873920142650604
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,64,power_law_1.01,0.040140798687934874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,64,power_law_1.01,0.6565248012542725
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,64,power_law_1.01,0.053491199016571046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,64,power_law_1.01,0.053420799970626834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,64,power_law_1.01,0.8584832191467285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,64,power_law_1.01,0.03991039991378784
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,64,power_law_1.01,0.039673599600791934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,64,power_law_1.01,0.03991680145263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,64,power_law_1.01,1.265113639831543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,64,power_law_1.01,0.039673599600791934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,64,power_law_1.01,0.041356799006462094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,64,power_law_1.01,0.043136000633239746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,64,power_law_1.01,1.6463552474975587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,64,power_law_1.01,3.6853759765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,64,power_law_1.2,0.038764798641204835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,64,power_law_1.2,0.03895680010318756
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,64,balanced,0.24502400557200113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,64,power_law_1.2,0.07123200297355652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,64,power_law_1.01,2.452467155456543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,64,balanced,0.34489599863688153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,64,power_law_1.2,0.07734400033950806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,64,balanced,0.447711984316508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,64,power_law_1.2,0.10115840435028076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,64,power_law_1.2,0.03887999951839447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,64,power_law_1.2,0.10176000595092774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,64,power_law_1.2,0.040575999021530154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,64,power_law_1.2,0.10613759756088256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,64,power_law_1.2,0.057625597715377806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,64,power_law_1.01,3.169331169128418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,64,power_law_1.2,0.06769919991493226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,64,power_law_1.2,0.05453439950942993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,64,power_law_1.2,0.09550719857215881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,64,power_law_1.2,0.05678079724311828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,64,power_law_1.2,0.09747200012207032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,64,power_law_1.2,1.191155242919922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,64,power_law_1.2,0.05711359977722168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,64,power_law_1.2,0.10039680004119873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,64,power_law_1.2,0.05882880091667175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,64,power_law_1.2,0.06875519752502442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,64,power_law_1.2,0.07004799842834472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,64,power_law_1.2,0.11021440029144287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,64,power_law_1.2,0.11871360540390015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,64,power_law_1.2,0.12291200160980224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,64,power_law_1.2,0.12878079414367677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,64,power_law_1.2,0.13362560272216797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,64,power_law_1.2,0.14001280069351196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,64,power_law_1.2,0.10265599489212036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,64,power_law_1.2,0.15139199495315553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,64,power_law_1.2,0.10602240562438965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,64,power_law_1.2,0.15991679430007935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,64,power_law_1.2,0.10769280195236205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,64,power_law_1.2,0.17251839637756347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,64,power_law_1.2,0.07246720194816589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,64,power_law_1.2,0.19093120098114014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,64,power_law_1.2,0.0753920018672943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,64,power_law_1.2,0.2383552074432373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,64,power_law_1.2,0.077811199426651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,64,power_law_1.2,0.07910400032997131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,64,power_law_1.2,0.08388479948043823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,64,power_law_1.2,0.09091200232505799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,64,power_law_1.2,0.0977728009223938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,64,power_law_1.2,0.10921599864959716
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,64,power_law_1.01,0.04464640021324158
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,64,power_law_1.2,0.11310080289840699
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,64,power_law_1.01,0.0500544011592865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,64,power_law_1.2,0.11983360052108764
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,64,power_law_1.01,0.056467199325561525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,64,power_law_1.2,0.12654080390930175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,64,power_law_1.2,0.2918528079986572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,64,power_law_1.2,0.13715840578079225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,64,power_law_1.2,0.14997119903564454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,64,power_law_1.2,0.4386688232421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,64,power_law_1.2,0.16277120113372803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,64,power_law_1.2,0.5356031894683838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,64,power_law_1.2,0.1870144009590149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,64,power_law_1.2,0.241593599319458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,64,power_law_1.2,0.12154879570007324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,64,power_law_1.2,0.29947519302368164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,64,power_law_1.2,0.13370239734649658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,64,power_law_1.2,0.43235201835632325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,64,power_law_1.2,0.1624575972557068
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,64,power_law_1.01,0.07536640167236328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,64,power_law_1.01,0.08888959884643555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,64,power_law_1.2,0.7811647891998291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,64,power_law_1.01,0.12276480197906495
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,64,power_law_1.01,0.15731199979782104
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,64,power_law_1.01,0.22143359184265138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,64,power_law_1.01,0.29487359523773193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,64,power_law_1.2,0.5510208129882812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,64,power_law_1.01,0.4059391975402832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,64,power_law_1.2,1.2397503852844238
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,64,power_law_1.01,0.5466815948486328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,64,power_law_1.2,0.7875391960144043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,64,power_law_1.2,1.2897343635559082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,64,power_law_1.2,0.21862399578094482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,64,power_law_1.2,0.989958381652832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,64,power_law_1.2,0.27370240688323977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,64,power_law_1.2,0.3796288013458252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,64,power_law_1.2,0.5060287952423096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,64,power_law_1.2,0.23968639373779296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,64,power_law_1.2,1.0700608253479005
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,64,power_law_1.2,0.024255999922752382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,64,power_law_1.2,0.02446720004081726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,64,power_law_1.2,1.5405759811401367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,64,power_law_1.2,0.024672000110149382
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,64,power_law_1.2,0.025465598702430724
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,64,power_law_1.2,0.027897599339485168
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,64,power_law_1.2,0.03112959861755371
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,64,power_law_1.2,0.038790398836135866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,64,power_law_1.2,0.6952832221984864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,64,power_law_1.2,0.040217599272727965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,64,power_law_1.2,0.05381119847297668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,64,power_law_1.2,0.05242239832878113
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,64,power_law_1.2,0.03918719887733459
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,64,power_law_1.2,0.03891839981079102
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,64,power_law_1.2,0.039724799990653994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,64,power_law_1.2,0.03909760117530823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,64,power_law_1.2,0.04202240109443665
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,64,power_law_1.2,0.04301440119743347
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,64,power_law_1.2,0.044998401403427125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,64,power_law_1.2,0.9386752128601075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,64,power_law_1.2,0.05375360250473023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,64,power_law_1.2,0.06111360192298889
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,64,power_law_1.2,0.08316159844398499
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,64,power_law_1.2,0.09271680116653443
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,64,power_law_1.2,0.13492480516433716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,64,power_law_1.2,1.5473407745361327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,64,power_law_1.2,0.16885119676589966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,64,power_law_1.2,1.3759552001953126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,64,power_law_1.2,1.6418432235717773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,64,balanced,0.0784853349129359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,64,balanced,0.07811200122038524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,64,balanced,0.07806399961312611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,64,balanced,0.1416106621424357
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,64,power_law_1.2,0.23145599365234376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,64,power_law_1.2,1.9481088638305664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,64,balanced,0.19559999306996664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,64,balanced,0.2743626634279887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,64,balanced,0.38126933574676514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,64,balanced,0.3843040068944295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,64,balanced,0.3876906633377075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,64,balanced,0.39375468095143634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,64,balanced,0.40161065260569256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,64,balanced,0.4084320068359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,64,balanced,0.416431983311971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,64,balanced,0.43025068442026776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,64,balanced,0.4469066858291626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,64,balanced,0.46190400918324787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,64,balanced,0.4943786859512329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,64,balanced,0.5544639825820923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,64,balanced,0.6145439942677816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,64,power_law_1.2,1.9321727752685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,64,balanced,0.7377280394236246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,64,balanced,0.857157309850057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,64,balanced,1.1029333273569744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,64,balanced,1.3978932698567708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,64,balanced,1.8788639704386394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,64,balanced,2.4467573165893555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,64,balanced,3.5224854151407876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,64,balanced,4.710778554280599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,64,balanced,0.04804266492525736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,64,balanced,0.056143999099731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,64,balanced,0.04978133241335551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,64,balanced,0.10027199983596802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,64,balanced,0.1620746652285258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,64,balanced,0.2627893288930257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,64,balanced,0.38317867120107013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,64,balanced,0.3842453161875407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,64,balanced,0.38976001739501953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,64,balanced,0.3937813440958659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,64,balanced,0.4029066562652588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,64,balanced,0.412224014600118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,64,balanced,0.4196053345998128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,64,balanced,0.4350666602452596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,64,balanced,0.45395731925964355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,64,balanced,0.4697599808375041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,64,balanced,0.5021706819534302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,64,balanced,0.563749353090922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,64,balanced,0.6305333375930786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,64,balanced,0.7547573248545328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,64,balanced,0.8808480103810629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,64,balanced,1.1351199944814045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,64,power_law_1.2,0.3331392049789429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,64,balanced,1.421845277150472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,64,balanced,1.9287734031677246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,64,power_law_1.2,1.8144319534301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,64,balanced,2.4925012588500977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,64,balanced,3.5340426762898765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,64,balanced,4.637071927388509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,64,balanced,0.07147199908892314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,64,balanced,0.07102400064468384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,64,balanced,0.07691200077533722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,64,balanced,0.07868800063927968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,64,balanced,0.11178666353225708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,64,balanced,0.16661866505940756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,64,balanced,0.2405653397242228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,64,balanced,0.24408000707626343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,64,balanced,0.24873065948486328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,64,balanced,0.2539680004119873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,64,balanced,0.2597493330637614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,64,balanced,0.2685973246892293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,64,balanced,0.2781546711921692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,64,balanced,0.2919306755065918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,64,balanced,0.3095253308614095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,64,balanced,0.3248639901479085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,64,balanced,0.35681601365407306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,64,balanced,0.4148906469345093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,64,balanced,0.4764053424199422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,64,balanced,0.5946880181630453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,64,balanced,0.7129759788513184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,64,balanced,0.9497439861297607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,64,balanced,1.221824010213216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,64,balanced,1.6916534105936687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,64,balanced,2.1947573026021323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,64,power_law_1.2,1.6938047409057617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,64,balanced,3.1718079249064126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,64,balanced,4.185381253560384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,64,power_law_1.01,0.12622720003128052
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,64,power_law_1.01,0.1296704053878784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,64,power_law_1.01,0.1313024044036865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,64,power_law_1.01,0.1762943983078003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,64,power_law_1.01,0.20736639499664306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,64,power_law_1.01,0.2783360004425049
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,64,power_law_1.2,0.3218816041946411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,64,power_law_1.01,0.10631680488586426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,64,power_law_1.01,0.3764031887054443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,64,power_law_1.01,0.10835200548171997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,64,power_law_1.01,0.3820352077484131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,64,power_law_1.01,0.38840959072113035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,64,balanced,0.0325546662012736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,64,balanced,0.03159466634194056
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,64,balanced,0.03162666658560435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,64,power_law_1.01,0.10856959819793702
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,64,balanced,0.03175999969244003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,64,balanced,0.033701332906881966
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,64,balanced,0.04385066529115041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,64,balanced,0.05843733251094818
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,64,balanced,0.06410666803518932
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,64,balanced,0.08518933256467183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,64,balanced,0.08656000097592671
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,64,balanced,0.2159199913342794
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,64,balanced,0.19503466288248697
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,64,balanced,0.18315200010935465
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,64,balanced,0.13900267084439596
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,64,balanced,0.08479467034339905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,64,balanced,0.08581866820653279
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,64,balanced,0.0902880032857259
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,64,balanced,0.09885866443316142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,64,balanced,0.10691199700037639
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,64,balanced,0.12327999869982402
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,64,balanced,0.14217600226402283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,64,balanced,0.18177600701649985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,64,balanced,0.22597867250442505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,64,balanced,0.3095093369483948
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,64,balanced,0.4013599952061971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,64,power_law_1.2,2.934073638916016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,64,power_law_1.01,0.391379189491272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,64,power_law_1.01,0.14123519659042358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,64,power_law_1.01,0.3958336114883423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,64,power_law_1.01,0.1766975998878479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,64,power_law_1.01,0.4066944122314453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,64,power_law_1.01,0.26410880088806155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,64,power_law_1.01,0.41349120140075685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,64,balanced,0.555621345837911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,64,power_law_1.01,0.37735040187835694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,64,power_law_1.01,0.4249983787536621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,64,balanced,0.7414240042368571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,64,power_law_1.01,0.44542717933654785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,64,power_law_1.01,0.4590144157409668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,64,power_law_1.01,0.4902656078338623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,64,power_law_1.01,0.5641344070434571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,64,power_law_1.01,0.6265920162200928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,64,power_law_1.01,0.3787584066390991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,64,power_law_1.01,0.7984511852264404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,64,power_law_1.01,0.38552958965301515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,64,power_law_1.01,0.9638912200927734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,64,power_law_1.01,0.39216639995574953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,64,power_law_1.2,2.5664960861206056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,64,power_law_1.01,0.40014081001281737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,64,power_law_1.01,0.40826878547668455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,64,power_law_1.01,0.10120320320129395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,64,power_law_1.01,0.10177279710769653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,64,power_law_1.01,0.10824960470199585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,64,power_law_1.01,0.418233585357666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,64,power_law_1.01,0.108787202835083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,64,power_law_1.01,0.4327040195465088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,64,power_law_1.01,1.2610688209533691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,64,power_law_1.01,0.12341760396957398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,64,power_law_1.01,0.4498432159423828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,64,power_law_1.01,0.16862080097198487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,64,power_law_1.01,0.46535677909851075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,64,power_law_1.01,0.23001599311828613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,64,power_law_1.01,0.5031871795654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,64,power_law_1.01,0.23877758979797364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,64,power_law_1.01,0.5769855976104736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,64,power_law_1.01,0.24170238971710206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,64,power_law_1.01,0.6511936187744141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,64,power_law_1.01,0.2495743989944458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,64,power_law_1.01,0.8159935951232911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,64,power_law_1.01,0.2573951959609985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,64,power_law_1.01,0.2615936040878296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,64,power_law_1.01,0.9556351661682129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,64,power_law_1.01,0.27082879543304444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,64,power_law_1.01,1.2809408187866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,64,power_law_1.01,0.2886912107467651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,64,power_law_1.2,2.9914176940917967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,64,power_law_1.01,0.3059200048446655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,64,power_law_1.01,1.5951552391052246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,64,power_law_1.01,0.31787519454956054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,64,power_law_1.01,1.578873634338379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,64,power_law_1.01,0.35218560695648193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,64,power_law_1.01,0.4208384037017822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,64,power_law_1.01,2.2068416595458986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,64,power_law_1.01,2.166387176513672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,64,power_law_1.01,0.4932415962219238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,64,power_law_1.01,0.6286272048950196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,64,power_law_1.01,2.8683263778686525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,64,power_law_1.01,2.8173952102661133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,64,power_law_1.01,0.768390417098999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,64,power_law_1.01,1.046560001373291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,64,power_law_1.01,1.302847957611084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,64,power_law_1.01,4.084076690673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,64,power_law_1.01,4.043718338012695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,64,power_law_1.01,1.8902080535888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,64,power_law_1.01,2.3967103958129883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,64,power_law_1.01,5.402617645263672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,64,power_law_1.01,3.550969696044922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,64,power_law_1.2,1.5611328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,64,power_law_1.01,5.2998912811279295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,64,power_law_1.2,0.12714240550994874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,64,power_law_1.2,0.1060479998588562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,64,power_law_1.2,0.13082879781723022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,64,power_law_1.2,0.1074560046195984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,64,power_law_1.2,0.1312000036239624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,64,power_law_1.01,4.6417793273925785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,64,power_law_1.2,0.10809600353240967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,64,power_law_1.2,0.1758784055709839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,64,power_law_1.2,0.10156799554824829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,64,power_law_1.2,0.14071040153503417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,64,power_law_1.2,0.20206079483032227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,64,power_law_1.2,0.10239360332489014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,64,power_law_1.2,0.17694720029830932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,64,power_law_1.2,0.27479679584503175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,64,power_law_1.2,0.10854400396347046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,64,power_law_1.2,0.26275200843811036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,64,power_law_1.2,0.10920319557189942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,64,power_law_1.2,0.37776000499725343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,64,power_law_1.2,0.125600004196167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,64,power_law_1.2,0.37921919822692873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,64,power_law_1.2,0.17140480279922485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,64,power_law_1.2,0.38575360774993894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,64,power_law_1.2,0.23146240711212157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,64,power_law_1.2,0.3909696102142334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,64,power_law_1.2,0.23427200317382812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,64,power_law_1.2,0.37531518936157227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,64,power_law_1.2,0.3972863912582397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,64,power_law_1.2,0.2450176000595093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,64,power_law_1.2,0.38278400897979736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,64,power_law_1.2,0.40921602249145506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,64,power_law_1.2,0.24837760925292968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,64,power_law_1.2,0.3865344047546387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,64,power_law_1.2,0.41541762351989747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,64,power_law_1.2,0.2536767959594727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,64,power_law_1.01,0.05075200200080872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,64,power_law_1.2,0.2636672019958496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,64,power_law_1.2,0.3869503974914551
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,64,power_law_1.01,0.050387197732925416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,64,power_law_1.2,0.2745599985122681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,64,power_law_1.2,0.3982784032821655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,64,power_law_1.01,0.051052802801132204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,64,power_law_1.2,0.28485119342803955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,64,power_law_1.2,0.4040832042694092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,64,power_law_1.01,0.050316798686981204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,64,power_law_1.2,0.30164480209350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,64,power_law_1.2,0.4133567810058594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,64,power_law_1.01,0.051046401262283325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,64,power_law_1.2,0.32117760181427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,64,power_law_1.2,0.4276991844177246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,64,power_law_1.01,0.05438719987869263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,64,power_law_1.2,0.35944960117340086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,64,power_law_1.2,0.4313920021057129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,64,power_law_1.2,0.4461184024810791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,64,power_law_1.01,0.05208960175514221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,64,power_law_1.2,0.4353472232818604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,64,power_law_1.2,0.45169920921325685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,64,power_law_1.2,0.4588160037994385
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,64,power_law_1.01,0.05611519813537598
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,64,power_law_1.2,0.5054336071014405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,64,power_law_1.2,0.4741824150085449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,64,power_law_1.2,0.4950272083282471
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,64,power_law_1.01,0.08251519799232483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,64,power_law_1.2,0.5069375991821289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,64,power_law_1.2,0.6385536193847656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,64,power_law_1.2,0.5681344032287597
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,64,power_law_1.01,0.07674880027770996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,64,power_law_1.2,0.5859583854675293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,64,power_law_1.2,0.7865856170654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,64,power_law_1.01,0.19440000057220458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,64,power_law_1.2,0.6642303943634034
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,64,power_law_1.01,0.1590783953666687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,64,power_law_1.2,1.1104640007019042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,64,power_law_1.2,0.8338239669799805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,64,power_law_1.01,0.1643455982208252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,64,power_law_1.2,1.0333696365356446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,64,power_law_1.01,0.08440960049629212
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,64,power_law_1.01,0.0807039976119995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,64,power_law_1.2,1.3354687690734863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,64,power_law_1.01,0.08183040022850037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,64,power_law_1.2,0.6645631790161133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,64,power_law_1.01,0.08823680281639099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,64,power_law_1.2,0.8162431716918945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,64,power_law_1.2,1.6726655960083008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,64,power_law_1.01,0.0887935996055603
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,64,power_law_1.01,0.09995520114898682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,64,power_law_1.2,1.0026047706604004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,64,power_law_1.2,1.3873727798461915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,64,power_law_1.01,0.12375040054321289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,64,power_law_1.2,2.4028480529785154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,64,power_law_1.2,1.3453184127807618
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,64,power_law_1.01,0.14515199661254882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,64,power_law_1.01,0.19222400188446045
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,64,power_law_1.01,0.24416000843048097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,64,power_law_1.2,1.7304512023925782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,64,power_law_1.2,3.2112449645996093
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,64,power_law_1.01,0.3444672107696533
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,64,power_law_1.01,0.4451712131500244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,64,power_law_1.2,2.352524757385254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,64,power_law_1.01,0.6321983814239502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,64,power_law_1.2,1.9436288833618165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,64,power_law_1.01,0.8764415740966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,64,power_law_1.2,3.0146240234375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,64,power_law_1.2,2.6367807388305664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,64,power_law_1.2,0.45542402267456056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,64,power_law_1.2,4.420800018310547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,64,power_law_1.2,4.580249786376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,64,power_law_1.2,3.7105857849121096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,64,power_law_1.2,6.19249267578125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,64,power_law_1.2,5.009759902954102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,64,balanced,0.033626665671666466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,64,balanced,0.03554133325815201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,64,balanced,0.03551999976237615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,64,balanced,0.03568533311287562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,64,balanced,0.05820266902446747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,64,balanced,0.08509866396586101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,64,balanced,0.1490239997704824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,64,balanced,0.1523306667804718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,64,balanced,0.15235199530919394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,64,balanced,0.15634666879971823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,64,balanced,0.1586666703224182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,64,balanced,0.16244799892107645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,64,balanced,0.16720000902811685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,64,balanced,0.17306667566299438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,64,balanced,0.18475200732549033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,64,balanced,0.19341333707173666
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,64,power_law_1.2,0.050348800420761106
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,64,power_law_1.2,0.050220799446105954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,64,power_law_1.2,6.142617416381836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,64,balanced,0.033557333052158356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,64,balanced,0.03347733368476232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,64,balanced,0.04884799818197886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,64,balanced,0.05807466804981232
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,64,power_law_1.2,0.0511680006980896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,64,balanced,0.049226666490236916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,64,balanced,0.08385066191355388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,64,balanced,0.05085866649945577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,64,balanced,0.14864533146222433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,64,balanced,0.053269331653912864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,64,balanced,0.15387733777364096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,64,balanced,0.06025066475073496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,64,balanced,0.09312533338864644
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,64,power_law_1.2,0.0500927984714508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,64,balanced,0.09727467099825542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,64,balanced,0.09900800387064616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,64,balanced,0.10108266274134318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,64,balanced,0.1034453312555949
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,64,power_law_1.2,0.05107839703559876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,64,balanced,0.10727999607721965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,64,balanced,0.1153546671072642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,64,balanced,0.11948800086975098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,64,balanced,0.1272426644961039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,64,balanced,0.1388800044854482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,64,balanced,0.2051253318786621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,64,balanced,0.14842666188875833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,64,balanced,0.22642133633295694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,64,balanced,0.26497066020965576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,64,balanced,0.305184006690979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,64,balanced,0.38521599769592285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,64,balanced,0.46559464931488037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,64,balanced,0.6239413420359293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,64,balanced,0.15482133626937866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,64,balanced,0.15665066242218018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,64,balanced,0.15879467129707336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,64,balanced,0.16448000073432922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,64,balanced,0.16858132680257162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,64,balanced,0.17476266622543335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,64,balanced,0.18338666359583536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,64,balanced,0.19323732455571493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,64,power_law_1.2,0.047353601455688475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,64,balanced,0.170799990495046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,64,balanced,0.20807466904322305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,64,power_law_1.2,0.055225598812103274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,64,balanced,0.2484266757965088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,64,balanced,0.3330880006154378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,64,balanced,0.4132373332977295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,64,power_law_1.2,0.060192000865936277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,64,balanced,0.5782986481984457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,64,balanced,0.7284426689147949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,64,balanced,0.7742933432261149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,64,balanced,1.0439626375834148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,64,balanced,1.0844213167826335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,64,balanced,0.2027626633644104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,64,balanced,1.3581600189208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,64,balanced,1.399818738301595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,64,balanced,1.9803519248962402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,64,balanced,2.039082686106364
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,64,power_law_1.2,0.07674239873886109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,64,balanced,2.607360045115153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,64,balanced,2.6621012687683105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,64,power_law_1.2,0.07090560197830201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,64,power_law_1.01,0.05716480016708374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,64,power_law_1.01,0.0590719997882843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,64,balanced,0.22388799985249838
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,64,power_law_1.2,0.16638720035552979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,64,balanced,0.2611626585324605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,64,balanced,0.30025599400202435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,64,balanced,0.3822773297627767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,64,balanced,0.45750399430592853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,64,balanced,0.6165759960810343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,64,balanced,0.7632373174031576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,64,balanced,1.068341334660848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,64,balanced,1.3690452575683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,64,power_law_1.01,0.057817602157592775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,64,balanced,0.026789332429567974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,64,balanced,0.0271519993742307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,64,power_law_1.2,0.15723520517349243
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,64,balanced,0.027434666951497395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,64,balanced,2.01910400390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,64,balanced,0.02738133321205775
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,64,balanced,0.029450667401154835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,64,balanced,0.03345066557327906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,64,balanced,0.035631999373435974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,64,power_law_1.2,0.15917439460754396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,64,balanced,0.036176001032193504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,64,balanced,0.043653334180514015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,64,balanced,0.04456000030040741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,64,power_law_1.2,3.4894142150878906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,64,balanced,0.06657066444555919
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,64,balanced,0.05801600217819214
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,64,balanced,0.05215999980767568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,64,power_law_1.01,0.04216319918632507
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,64,balanced,0.04586133360862732
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,64,balanced,0.045781334241231285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,64,balanced,0.04784533381462097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,64,power_law_1.01,0.04176000058650971
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,64,balanced,0.047882666190465294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,64,balanced,0.0484799991051356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,64,balanced,0.050213331977526345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,64,power_law_1.01,0.0424703985452652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,64,balanced,0.0599839985370636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,64,balanced,0.0684799998998642
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,64,balanced,0.08879466851552327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,64,power_law_1.01,0.05889279842376709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,64,power_law_1.01,0.0871936023235321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,64,power_law_1.01,0.08600320219993592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,64,power_law_1.01,0.0932416021823883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,64,power_law_1.01,0.14640640020370482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,64,power_law_1.01,0.14791040420532225
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,64,power_law_1.2,0.10898560285568237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,64,power_law_1.01,0.1490623950958252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,64,power_law_1.01,0.15113600492477416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,64,power_law_1.2,0.08232319951057435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,64,power_law_1.01,0.15301120281219482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,64,power_law_1.2,0.0759552001953125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,64,power_law_1.01,0.15531519651412964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,64,power_law_1.01,0.15709439516067505
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,64,balanced,0.11361066500345866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,64,power_law_1.01,0.1614848017692566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,64,power_law_1.01,0.166758394241333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,64,power_law_1.01,0.1717568039894104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,64,power_law_1.01,0.18166400194168092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,64,power_law_1.01,0.19187840223312377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,64,power_law_1.2,3.964543914794922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,64,power_law_1.01,0.20317440032958983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,64,power_law_1.01,0.05079039931297302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,64,power_law_1.01,0.05089920163154602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,64,power_law_1.01,0.055219197273254396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,64,balanced,2.640906651814779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,64,power_law_1.01,0.05541759729385376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,64,power_law_1.01,0.061343997716903687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,64,power_law_1.01,0.15100799798965453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,64,power_law_1.01,0.09162880182266235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,64,power_law_1.01,0.15454720258712767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,64,power_law_1.01,0.09475839734077454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,64,power_law_1.01,0.22312960624694825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,64,power_law_1.01,0.15541759729385377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,64,power_law_1.01,0.09630079865455628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,64,power_law_1.01,0.263155198097229
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,64,balanced,0.16054399808247885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,64,balanced,0.20247467358907065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,64,power_law_1.01,0.09905279874801635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,64,power_law_1.01,0.30256640911102295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,64,balanced,0.2863626678784688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,64,power_law_1.01,0.10065920352935791
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,64,balanced,0.3696266810099284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,64,power_law_1.01,0.10565760135650634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,64,power_law_1.2,0.05687040090560913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,64,power_law_1.01,0.11173759698867798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,64,power_law_1.2,0.05661439895629883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,64,power_law_1.01,0.11667200326919555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,64,power_law_1.2,0.05806080102920532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,64,power_law_1.01,0.12617599964141846
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,64,power_law_1.01,0.026361599564552307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,64,power_law_1.01,0.137990403175354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,64,power_law_1.01,0.1610368013381958
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,64,power_law_1.01,0.025996801257133485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,64,power_law_1.01,0.14945919513702394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,64,power_law_1.01,0.16607359647750855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,64,power_law_1.01,0.38966400623321534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,64,power_law_1.01,0.17129600048065186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,64,power_law_1.01,0.1713088035583496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,64,power_law_1.01,0.47387518882751467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,64,power_law_1.01,0.18253439664840698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,64,power_law_1.01,0.1938367962837219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,64,power_law_1.01,0.6585279941558838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,64,power_law_1.01,0.20478079319000245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,64,power_law_1.2,0.08216959834098816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,64,power_law_1.01,0.22441599369049073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,64,power_law_1.01,0.8360768318176269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,64,power_law_1.2,0.09432960152626038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,64,power_law_1.01,0.2654911994934082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,64,power_law_1.2,0.14723199605941772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,64,power_law_1.01,1.1751615524291992
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,64,power_law_1.01,0.026982399821281432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,64,power_law_1.2,0.15060479640960694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,64,power_law_1.01,0.21322240829467773
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,64,power_law_1.01,0.027385601401329042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,64,power_law_1.2,0.15181440114974976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,64,power_law_1.01,1.5344575881958007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,64,power_law_1.01,0.2605760097503662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,64,power_law_1.01,0.02813439965248108
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,64,power_law_1.2,0.08170239925384522
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,64,power_law_1.01,0.031020799279212953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,64,power_law_1.2,0.09301120042800903
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,64,power_law_1.01,0.03343360126018524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,64,power_law_1.01,2.2102592468261717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,64,power_law_1.2,0.10480639934539795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,64,power_law_1.01,0.0342848002910614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,64,power_law_1.01,0.3049407958984375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,64,power_law_1.2,0.125164794921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,64,power_law_1.01,0.04169600009918213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,64,power_law_1.01,0.3955712080001831
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,64,power_law_1.01,0.041843199729919435
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,64,power_law_1.2,0.14890240430831908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,64,power_law_1.01,0.4888063907623291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,64,power_law_1.01,0.06047999858856201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,64,power_law_1.2,0.15463680028915405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,64,power_law_1.2,0.2010432004928589
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,64,power_law_1.01,0.052211201190948485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,64,power_law_1.01,0.684819221496582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,64,power_law_1.2,0.2591295957565308
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,64,power_law_1.01,0.04850560128688812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,64,power_law_1.01,0.8432703971862793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,64,power_law_1.2,0.3761600017547607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,64,power_law_1.2,3.874879837036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,64,power_law_1.01,1.2019776344299316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,64,power_law_1.2,0.48684158325195315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,64,power_law_1.2,0.04059520065784454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,64,power_law_1.2,0.041817599534988405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,64,power_law_1.2,0.6881663799285889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,64,power_law_1.01,1.5434368133544922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,64,power_law_1.01,0.3475136041641235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,64,power_law_1.2,0.15656319856643677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,64,power_law_1.01,0.4413311958312988
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,64,power_law_1.01,0.042508798837661746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,64,power_law_1.2,0.16090240478515624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,64,power_law_1.01,0.6160255908966065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,64,power_law_1.01,2.271993637084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,64,power_law_1.01,0.04286080002784729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,64,power_law_1.2,0.16661759614944457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,64,power_law_1.01,0.04552960097789764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,64,power_law_1.2,0.17095680236816407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,64,power_law_1.01,0.779744005203247
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,64,power_law_1.01,0.04496000111103058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,64,power_law_1.2,0.18079999685287476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,64,power_law_1.01,2.927084732055664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,64,power_law_1.01,0.047167998552322385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,64,power_law_1.01,1.1106752395629882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,64,power_law_1.01,0.05252479910850525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,64,power_law_1.01,2.9526399612426757
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,64,power_law_1.01,0.06752640008926392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,64,power_law_1.2,0.9490431785583496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,64,power_law_1.01,1.4498304367065429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,64,balanced,0.03367999941110611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,64,balanced,0.033589333295822144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,64,balanced,0.03364266703526179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,64,balanced,0.05622399846712748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,64,balanced,0.08273600041866302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,64,balanced,0.15012266238530478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,64,balanced,0.15430933237075806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,64,balanced,0.15424000223477682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,64,balanced,0.15845333536465964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,64,power_law_1.01,2.143654441833496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,64,balanced,0.15939733386039734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,64,balanced,0.16327466567357382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,64,power_law_1.2,0.1922943949699402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,64,power_law_1.2,0.20205440521240234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,64,power_law_1.2,0.050457602739334105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,64,power_law_1.2,0.22172160148620607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,64,power_law_1.2,0.051635199785232545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,64,power_law_1.2,0.04197759926319122
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,64,power_law_1.01,0.07587199807167053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,64,power_law_1.2,0.05555840134620667
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,64,power_law_1.01,0.10234240293502808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,64,power_law_1.2,0.055276799201965335
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,64,power_law_1.01,0.128601598739624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,64,power_law_1.2,0.06056960225105286
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,64,power_law_1.01,0.17812479734420777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,64,power_law_1.2,0.09168639779090881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,64,balanced,0.16873600085576376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,64,balanced,0.1727893352508545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,64,balanced,0.18422933419545492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,64,power_law_1.2,0.09556480050086975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,64,balanced,0.1950613260269165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,64,balanced,0.20331199963887533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,64,power_law_1.2,0.09678080081939697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,64,balanced,0.2237493395805359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,64,balanced,0.2608106732368469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,64,power_law_1.2,0.2642944097518921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,64,power_law_1.2,0.0996288001537323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,64,balanced,0.2998879949251811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,64,power_law_1.2,0.05888640284538269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,64,power_law_1.2,0.30674560070037843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,64,power_law_1.2,0.08675839900970458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,64,power_law_1.2,0.3928767919540405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,64,power_law_1.2,0.02568959891796112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,64,power_law_1.2,0.14664959907531738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,64,power_law_1.2,0.5089280128479003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,64,power_law_1.2,0.025331199169158936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,64,power_law_1.01,0.22943360805511476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,64,power_law_1.2,0.15048960447311402
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,64,power_law_1.2,0.026873600482940675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,64,power_law_1.2,0.6771071910858154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,64,power_law_1.01,0.32355198860168455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,64,power_law_1.2,0.15172480344772338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,64,power_law_1.2,0.027008000016212463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,64,power_law_1.2,0.853228759765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,64,power_law_1.2,0.15520000457763672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,64,power_law_1.2,0.027398398518562316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,64,balanced,0.381930669148763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,64,power_law_1.2,0.15702400207519532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,64,power_law_1.2,0.10051200389862061
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,64,power_law_1.2,0.03165439963340759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,64,balanced,0.4596800009409587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,64,power_law_1.2,0.16085120439529418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,64,power_law_1.2,0.10644479990005493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,64,balanced,0.6145653327306112
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,64,power_law_1.2,0.03357439935207367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,64,power_law_1.2,0.16702719926834106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,64,power_law_1.2,0.11143679618835449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,64,balanced,0.7623519897460938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,64,power_law_1.2,0.03434880077838898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,64,power_law_1.2,0.17151360511779784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,64,power_law_1.2,0.11598080396652222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,64,balanced,1.0626719792683919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,64,power_law_1.2,0.18250240087509156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,64,power_law_1.2,0.041152000427246094
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,64,power_law_1.2,0.6235007762908935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,64,power_law_1.2,0.04162560105323791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,64,balanced,0.03363200028737386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,64,balanced,0.03540800015131632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,64,balanced,0.03359466542800268
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,64,power_law_1.01,0.43980798721313474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,64,power_law_1.2,0.057792001962661745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,64,power_law_1.01,2.8165119171142576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,64,balanced,0.048858667413393654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,64,balanced,0.04789333542188009
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,64,power_law_1.2,0.05018879771232605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,64,balanced,0.05314133564631144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,64,balanced,0.05285866558551788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,64,balanced,0.060319999853769936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,64,power_law_1.2,1.2657855987548827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,64,balanced,0.09289066990216573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,64,balanced,0.09521599610646565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,64,balanced,0.09916800260543823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,64,balanced,0.10122133294741313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,64,balanced,0.10459733009338379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,64,balanced,0.10789333780606587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,64,balanced,0.11402133107185364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,64,power_law_1.2,1.6652992248535157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,64,balanced,0.11753066380818684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,64,balanced,0.1297706663608551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,64,power_law_1.2,0.12817920446395875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,64,balanced,0.14205333590507507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,64,balanced,0.14828800161679587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,64,power_law_1.2,0.14039039611816406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,64,balanced,1.3670345942179363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,64,power_law_1.2,0.1519871950149536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,64,balanced,0.058415999015172325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,64,balanced,0.08483200271924336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,64,balanced,0.1483733355998993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,64,power_law_1.2,0.17605760097503662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,64,balanced,0.15249066551526388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,64,balanced,2.012597401936849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,64,balanced,0.15238400300343832
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,64,power_law_1.2,0.04746240079402923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,64,power_law_1.2,0.219596791267395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,64,power_law_1.2,0.042284798622131345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,64,power_law_1.2,0.26099839210510256
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,64,balanced,0.02510400116443634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,64,balanced,0.025498665869235992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,64,balanced,2.6302666664123535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,64,balanced,0.0272533322374026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,64,power_law_1.2,0.3610368013381958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,64,power_law_1.01,0.057792001962661745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,64,power_law_1.2,0.1948415994644165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,64,power_law_1.2,0.45055360794067384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,64,balanced,0.1695093313852946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,64,balanced,0.2095573345820109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,64,power_law_1.2,0.2050623893737793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,64,balanced,0.25125332673390705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,64,power_law_1.2,0.6223423957824707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,64,balanced,0.33240000406901044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,64,power_law_1.2,2.4153984069824217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,64,balanced,0.41331199804941815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,64,balanced,0.15668267011642456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,64,balanced,0.5791626771291097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,64,balanced,0.1586133340994517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,64,balanced,0.16266133387883505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,64,balanced,0.7312266826629639
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,64,power_law_1.2,0.042656001448631284
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,64,balanced,0.027242665489514668
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,64,balanced,0.029167999823888142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,64,power_law_1.2,0.04433279931545257
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,64,balanced,0.033333333830038704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,64,power_law_1.2,3.2180095672607423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,64,balanced,0.03502399971087774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,64,balanced,0.035386666655540466
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,64,power_law_1.2,0.045638400316238406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,64,balanced,0.04359999795754751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,64,power_law_1.01,0.042387199401855466
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,64,balanced,0.04389866689840952
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,64,balanced,0.0661653329928716
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,64,balanced,0.05589866638183594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,64,power_law_1.2,0.22673919200897216
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,64,balanced,0.05172266562779745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,64,balanced,0.044863998889923096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,64,balanced,0.045647998650868736
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,64,balanced,0.04804799954096476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,64,power_law_1.2,0.7989952087402343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,64,balanced,0.047781333327293396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,64,balanced,0.04809600114822388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,64,balanced,0.16843199729919434
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,64,balanced,0.049914668003718056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,64,balanced,0.17281599839528403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,64,balanced,0.05997333427270254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,64,balanced,0.1834826668103536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,64,balanced,0.06842133402824402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,64,power_law_1.2,1.1902015686035157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,64,balanced,0.08752000331878662
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,64,balanced,0.1132586697737376
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,64,balanced,0.16051200032234192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,64,power_law_1.01,0.0582144021987915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,64,balanced,0.20137067635854086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,64,power_law_1.2,0.04828799962997436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,64,power_law_1.01,0.05761280059814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,64,power_law_1.2,1.5658111572265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,64,power_law_1.01,0.042131200432777405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,64,power_law_1.2,0.05729280114173889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,64,power_law_1.01,0.08568320274353028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,64,power_law_1.01,0.04267520010471344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,64,power_law_1.2,0.2688447952270508
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,64,power_law_1.2,0.06936960220336914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,64,power_law_1.01,0.0597055971622467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,64,power_law_1.2,0.3116352081298828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,64,power_law_1.2,0.07894399762153625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,64,power_law_1.2,2.1985599517822267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,64,power_law_1.01,0.08574079871177673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,64,power_law_1.2,0.40881919860839844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,64,balanced,1.0463199615478516
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,64,power_law_1.2,0.10588799715042115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,64,power_law_1.01,0.14703359603881835
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,64,power_law_1.2,0.13175040483474731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,64,power_law_1.2,0.5277376174926758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,64,balanced,1.3625599543253581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,64,power_law_1.01,0.14952319860458374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,64,power_law_1.2,0.19276800155639648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,64,balanced,0.28709866603215534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,64,power_law_1.01,0.15148160457611085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,64,power_law_1.2,0.6936831951141358
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,64,balanced,0.36981332302093506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,64,balanced,1.9871519406636555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,64,power_law_1.01,0.15420160293579102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,64,power_law_1.01,0.05050240159034729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,64,power_law_1.2,0.9173952102661133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,64,power_law_1.01,0.15556479692459108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,64,power_law_1.01,0.05095679759979248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,64,power_law_1.01,0.16158080101013184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,64,balanced,2.609114646911621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,64,power_law_1.01,0.05568640232086182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,64,power_law_1.2,1.2493632316589356
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,64,power_law_1.01,0.16632959842681885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,64,power_law_1.01,0.05614079833030701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,64,power_law_1.01,0.1713536024093628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,64,power_law_1.01,0.18197120428085328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,64,power_law_1.2,1.648396873474121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,64,power_law_1.01,0.19256319999694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,64,power_law_1.2,2.9765119552612305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,64,power_law_1.01,0.20107519626617432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,64,power_law_1.2,0.05788159966468811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,64,power_law_1.01,0.22334721088409423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,64,power_law_1.01,0.09284480214118958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,64,power_law_1.2,0.057120001316070555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,64,power_law_1.01,0.2650559902191162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,64,power_law_1.2,2.455046463012695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,64,power_law_1.01,0.14880000352859496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,64,power_law_1.2,0.05800960063934326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,64,power_law_1.01,0.3019968032836914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,64,balanced,0.19738666216532388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,64,power_law_1.01,0.15105279684066772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,64,balanced,0.20572799444198608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,64,power_law_1.2,0.0840511977672577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,64,balanced,0.22666666905085245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,64,power_law_1.01,0.3876607894897461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,64,balanced,0.26580266157786053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,64,power_law_1.2,0.09290879964828491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,64,balanced,0.30776000022888184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,64,power_law_1.01,0.47925758361816406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,64,power_law_1.2,0.1469375967979431
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,64,power_law_1.2,3.292192077636719
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,64,power_law_1.2,0.24771199226379395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,64,power_law_1.01,0.6708735942840576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,64,power_law_1.2,0.15049599409103392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,64,power_law_1.2,0.041126400232315063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,64,power_law_1.2,0.15318399667739868
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,64,power_law_1.2,0.3686079978942871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,64,power_law_1.01,0.8262271881103516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,64,power_law_1.2,0.04192639887332916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,64,power_law_1.2,0.15523200035095214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,64,power_law_1.2,0.041715198755264284
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,64,power_law_1.2,0.47081599235534666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,64,power_law_1.2,0.15696640014648439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,64,power_law_1.01,1.2002816200256348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,64,power_law_1.2,0.05059199929237366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,64,power_law_1.01,0.0615231990814209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,64,balanced,0.38809065024058026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,64,power_law_1.01,0.09192320108413696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,64,power_law_1.01,1.5340415954589843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,64,power_law_1.01,0.026092800498008727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,64,power_law_1.01,0.09526399970054626
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,64,power_law_1.01,0.025939199328422546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,64,power_law_1.01,0.09696000218391418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,64,power_law_1.01,0.027046400308609008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,64,power_law_1.01,0.10013439655303955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,64,power_law_1.01,2.2353984832763674
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,64,power_law_1.01,0.027398398518562316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,64,power_law_1.2,0.05916799902915955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,64,power_law_1.01,0.02826879918575287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,64,power_law_1.2,0.086080002784729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,64,power_law_1.2,0.0514303982257843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,64,power_law_1.01,0.031436800956726074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,64,balanced,0.4702933231989543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,64,power_law_1.2,0.1472640037536621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,64,power_law_1.2,0.05533440113067627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,64,power_law_1.01,0.03353599905967712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,64,power_law_1.01,2.9625408172607424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,64,balanced,0.6306346654891968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,64,power_law_1.2,0.15013760328292847
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,64,power_law_1.01,0.03466239869594574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,64,balanced,0.7842240333557129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,64,power_law_1.2,0.1525760054588318
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,64,power_law_1.01,0.041791999340057374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,64,balanced,1.0975039800008137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,64,power_law_1.2,0.15509120225906373
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,64,power_law_1.01,0.04170880019664765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,64,power_law_1.01,0.1012992024421692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,64,balanced,1.4129652976989746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,64,power_law_1.2,0.1572543978691101
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,64,power_law_1.01,0.061766397953033444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,64,power_law_1.01,0.1069375991821289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,64,power_law_1.2,0.1622015953063965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,64,power_law_1.01,0.04874880015850067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,64,balanced,2.0583680470784507
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,64,power_law_1.01,0.04924800097942352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,64,power_law_1.01,0.11178879737854004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,64,power_law_1.2,0.05586559772491455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,64,power_law_1.01,0.11534719467163086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,64,balanced,2.6906506220499673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,64,power_law_1.2,0.06074879765510559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,64,power_law_1.01,0.1265663981437683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,64,balanced,0.03349333256483078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,64,balanced,0.03367999941110611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,64,balanced,0.053472002347310386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,64,balanced,0.05899199843406677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,64,balanced,0.07040533423423767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,64,balanced,0.08787199854850769
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,64,power_law_1.2,0.026054400205612182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,64,power_law_1.2,0.16223360300064088
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,64,power_law_1.2,0.025593599677085875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,64,power_law_1.2,0.1676543951034546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,64,power_law_1.2,0.026943999528884887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,64,power_law_1.2,0.17301119565963746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,64,power_law_1.01,0.04163840115070343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,64,power_law_1.2,0.02682879865169525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,64,power_law_1.2,0.18330880403518676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,64,power_law_1.01,0.04417920112609863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,64,power_law_1.2,0.027884799242019653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,64,power_law_1.2,0.19359359741210938
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,64,power_law_1.01,0.04493440091609955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,64,power_law_1.2,0.030995199084281923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,64,power_law_1.2,0.2055743932723999
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,64,power_law_1.01,0.04504320025444031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,64,power_law_1.2,0.22855041027069092
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,64,power_law_1.2,0.03367680013179779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,64,power_law_1.01,0.15292160511016845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,64,power_law_1.2,0.26988160610198975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,64,power_law_1.2,0.03451519906520843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,64,power_law_1.01,0.15537919998168945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,64,power_law_1.2,0.31176319122314455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,64,power_law_1.2,0.04168320000171662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,64,power_law_1.2,0.4040639877319336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,64,power_law_1.2,0.5077568054199219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,64,power_law_1.01,0.13925119638442993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,64,power_law_1.01,0.14985599517822265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,64,balanced,0.08896533648173015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,64,balanced,0.09087466200192769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,64,power_law_1.01,0.1716223955154419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,64,power_law_1.2,0.16740479469299316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,64,power_law_1.01,0.2153856039047241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,64,power_law_1.2,0.17142399549484252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,64,power_law_1.01,0.15758719444274902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,64,power_law_1.01,0.26139519214630125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,64,power_law_1.2,0.04136320054531097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,64,power_law_1.01,0.3512192010879517
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,64,power_law_1.2,0.05990399718284607
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,64,power_law_1.2,0.053830397129058835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,64,power_law_1.01,0.44141440391540526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,64,power_law_1.2,0.6997056007385254
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,64,power_law_1.2,0.04604800045490265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,64,power_law_1.01,0.04690560102462769
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,64,power_law_1.2,0.043609601259231565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,64,power_law_1.2,0.8863424301147461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,64,balanced,0.09125866492589314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,64,power_law_1.01,0.052960002422332765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,64,power_law_1.2,0.04261119961738587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,64,balanced,0.09300800164540608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,64,balanced,0.09707732995351155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,64,balanced,0.10087999701499939
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,64,power_law_1.01,0.06748800277709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,64,power_law_1.2,0.04364160001277924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,64,balanced,0.10334933797518413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,64,balanced,0.11156266927719116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,64,balanced,0.11961600184440613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,64,power_law_1.2,1.267801570892334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,64,balanced,0.12825066844622293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,64,balanced,0.14512000481287637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,64,balanced,0.1748746633529663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,64,balanced,0.20459200938542685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,64,balanced,0.26785600185394287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,64,balanced,0.33162132898966473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,64,power_law_1.2,1.6484031677246094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,64,power_law_1.2,0.0906175971031189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,64,power_law_1.01,0.6145984172821045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,64,power_law_1.2,0.09492480158805847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,64,power_law_1.2,0.09655680060386658
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,64,power_law_1.01,0.778656005859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,64,power_law_1.2,2.42989444732666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,64,power_law_1.2,0.09946879744529724
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,64,power_law_1.01,0.07736319899559022
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,64,power_law_1.2,0.04537599980831146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,64,power_law_1.01,0.16251519918441773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,64,power_law_1.01,1.1071871757507323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,64,power_law_1.2,0.10067839622497558
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,64,power_law_1.2,0.049798399209976196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,64,power_law_1.01,0.1674623966217041
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,64,power_law_1.2,0.054655998945236206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,64,power_law_1.01,0.1721343994140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,64,balanced,0.46448532740275067
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,64,power_law_1.2,0.06652160286903382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,64,power_law_1.01,0.1817919969558716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,64,power_law_1.01,1.4387007713317872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,64,power_law_1.2,3.140825653076172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,64,balanced,0.5860639810562134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,64,balanced,0.033914667864640556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,64,power_law_1.2,0.08178560137748718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,64,power_law_1.01,0.19247360229492189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,64,balanced,0.8367679913838705
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,64,power_law_1.2,0.109990394115448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,64,balanced,1.0852213700612385
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,64,power_law_1.2,0.13928960561752318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,64,power_law_1.01,2.1199615478515623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,64,power_law_1.2,0.18217600584030152
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,64,power_law_1.2,0.18824319839477538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,64,balanced,1.5970346132914226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,64,power_law_1.2,0.19329919815063476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,64,power_law_1.2,0.24819839000701904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,64,power_law_1.2,0.20268800258636474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,64,power_law_1.2,0.34990079402923585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,64,power_law_1.01,2.7924095153808595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,64,balanced,0.033999999364217125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,64,power_law_1.2,0.4598976135253906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,64,power_law_1.01,0.033932799100875856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,64,power_law_1.01,0.20251519680023194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,64,power_law_1.01,0.03383679986000061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,64,power_law_1.01,0.03473280072212219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,64,power_law_1.01,0.22220160961151122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,64,power_law_1.01,0.055174398422241214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,64,power_law_1.01,0.055795198678970336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,64,power_law_1.01,0.1044543981552124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,64,power_law_1.01,0.0688256025314331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,64,power_law_1.2,0.10560640096664428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,64,power_law_1.01,0.08538879752159119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,64,power_law_1.01,0.0864960014820099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,64,power_law_1.2,0.22310400009155273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,64,power_law_1.01,0.08913279771804809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,64,power_law_1.2,0.26347520351409914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,64,power_law_1.01,0.09042559862136841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,64,balanced,0.05215999980767568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,64,balanced,0.05205333232879639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,64,power_law_1.01,0.09095039963722229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,64,power_law_1.2,0.30709118843078614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,64,balanced,0.07245866457621257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,64,balanced,0.0888320008913676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,64,power_law_1.01,0.0953279972076416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,64,power_law_1.2,0.3999232053756714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,64,power_law_1.01,0.03621760010719299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,64,power_law_1.01,0.09991040229797363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,64,power_law_1.01,0.2632960081100464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,64,power_law_1.01,0.1019711971282959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,64,power_law_1.01,0.050892800092697144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,64,power_law_1.2,0.4885119915008545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,64,power_law_1.01,0.11187200546264649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,64,power_law_1.01,0.05118079781532288
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,64,power_law_1.01,0.13071999549865723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,64,power_law_1.01,0.12001919746398926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,64,power_law_1.01,0.07170559763908387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,64,power_law_1.2,0.6816895961761474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,64,power_law_1.2,0.11185920238494873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,64,balanced,2.128122647603353
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,64,power_law_1.01,0.1784832000732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,64,power_law_1.01,0.08788480162620545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,64,power_law_1.2,0.11664639711380005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,64,power_law_1.2,0.0344895988702774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,64,power_law_1.01,0.23066880702972412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,64,power_law_1.2,0.12728960514068605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,64,power_law_1.2,0.14113919734954833
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,64,power_law_1.01,0.33045759201049807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,64,power_law_1.2,0.15137280225753785
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,64,power_law_1.01,0.43258237838745117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,64,power_law_1.2,0.1719488024711609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,64,power_law_1.2,0.034643200039863584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,64,power_law_1.01,0.30388479232788085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,64,power_law_1.2,0.22013440132141113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,64,power_law_1.2,0.03619840145111084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,64,power_law_1.01,0.3917056083679199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,64,power_law_1.01,0.1280128002166748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,64,power_law_1.2,0.05021439790725708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,64,power_law_1.01,0.14457600116729735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,64,power_law_1.2,0.05143679976463318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,64,power_law_1.2,0.8976896286010743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,64,power_law_1.01,0.17811839580535888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,64,power_law_1.2,0.03314560055732727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,64,power_law_1.2,0.07103999853134155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,64,power_law_1.2,0.056383997201919556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,64,power_law_1.2,0.08759679794311523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,64,balanced,0.09148800373077393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,64,balanced,0.09333866834640503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,64,power_law_1.2,0.05413119792938233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,64,power_law_1.2,0.08913919925689698
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,64,power_law_1.2,1.250227165222168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,64,power_law_1.2,0.0916159987449646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,64,power_law_1.2,0.06872959733009339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,64,power_law_1.2,0.09317759871482849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,64,power_law_1.2,0.08520960211753845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,64,power_law_1.2,0.09448959827423095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,64,power_law_1.2,0.08635519742965699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,64,power_law_1.2,1.6766975402832032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,64,power_law_1.2,0.09807999730110169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,64,power_law_1.2,0.08973439931869506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,64,power_law_1.2,0.2677119970321655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,64,power_law_1.2,0.10217599868774414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,64,power_law_1.2,0.09095039963722229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,64,power_law_1.01,0.48021759986877444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,64,power_law_1.2,0.10577919483184814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,64,power_law_1.2,0.357587194442749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,64,power_law_1.2,0.11393920183181763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,64,power_law_1.01,0.6517824172973633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,64,power_law_1.2,2.353049659729004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,64,power_law_1.2,0.4570303916931152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,64,balanced,0.09342400232950847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,64,balanced,0.09549867113431294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,64,balanced,0.09909333785374959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,64,power_law_1.2,0.6280447959899902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,64,balanced,0.10520000259081523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,64,balanced,0.10757866501808167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,64,power_law_1.01,0.08926720023155213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,64,balanced,0.11755733688672383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,64,power_law_1.2,0.09142400026321411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,64,balanced,0.12171733379364014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,64,power_law_1.2,0.8332480430603028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,64,balanced,0.12990400195121765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,64,power_law_1.2,0.095551997423172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,64,balanced,0.14643200238545737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,64,power_law_1.01,0.21086719036102294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,64,balanced,0.18094400564829508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,64,power_law_1.2,0.12342400550842285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,64,power_law_1.2,0.09953920245170593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,64,power_law_1.2,3.098182487487793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,64,balanced,0.21241066853205362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,64,power_law_1.01,0.2788288116455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,64,balanced,0.04375466704368591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,64,power_law_1.2,0.10251519680023194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,64,power_law_1.2,1.1562175750732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,64,power_law_1.01,0.8394432067871094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,64,power_law_1.2,0.11400320529937744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,64,power_law_1.2,0.12135679721832275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,64,power_law_1.01,1.1972991943359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,64,power_law_1.2,0.12973439693450928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,64,power_law_1.2,1.5043392181396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,64,power_law_1.2,0.14664959907531738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,64,power_law_1.01,0.09188479781150818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,64,power_law_1.2,0.17929600477218627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,64,power_law_1.01,0.09303039908409119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,64,power_law_1.01,1.523526382446289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,64,power_law_1.01,0.09435520172119141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,64,power_law_1.2,2.267622375488281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,64,balanced,0.07514666517575581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,64,power_law_1.01,0.09861760139465332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,64,balanced,0.08365333080291748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,64,power_law_1.01,0.35630719661712645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,64,power_law_1.01,0.10234240293502808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,64,power_law_1.01,0.10522240400314331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,64,power_law_1.01,0.5134848117828369
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,64,power_law_1.01,2.1926912307739257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,64,power_law_1.01,0.1149440050125122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,64,power_law_1.01,0.12172800302505493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,64,power_law_1.01,0.6658815860748291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,64,power_law_1.2,2.9870464324951174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,64,power_law_1.01,0.1289728045463562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,64,balanced,0.04580800235271454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,64,balanced,0.048063998421033226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,64,balanced,0.0689333329598109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,64,power_law_1.01,0.9672127723693847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,64,balanced,0.09126399954160054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,64,balanced,0.30807467301686603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,64,power_law_1.01,2.9174144744873045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,64,balanced,0.4307253360748291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,64,balanced,0.4386719862620036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,64,balanced,0.06551999847094218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,64,power_law_1.01,1.2692480087280273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,64,balanced,0.07266666491826375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,64,balanced,0.4475413163503011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,64,balanced,0.08308266599973042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,64,balanced,0.12780800461769104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,64,balanced,0.45732800165812176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,64,balanced,0.20970133940378824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,64,balanced,0.4651679992675781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,64,balanced,0.2737706700960795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,64,balanced,0.2834986646970113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,64,balanced,0.29384533564249676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,64,power_law_1.01,1.8892736434936523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,64,balanced,0.3019253412882487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,64,balanced,0.31172800064086914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,64,power_law_1.01,0.14644479751586914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,64,power_law_1.01,0.179366397857666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,64,power_law_1.2,0.21390080451965332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,64,power_law_1.2,0.13169280290603638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,64,power_law_1.01,2.497580718994141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,64,power_law_1.2,0.1484992027282715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,64,balanced,0.10895466804504395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,64,balanced,0.307914674282074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,64,power_law_1.2,0.1832128047943115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,64,power_law_1.01,0.12565759420394898
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,64,balanced,0.4310186703999837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,64,power_law_1.2,0.22052481174468994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,64,balanced,0.48206933339436847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,64,power_law_1.2,0.29687678813934326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,64,balanced,0.49956266085306805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,64,balanced,0.5228480100631714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,64,power_law_1.2,0.38431999683380125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,64,balanced,0.3285920023918152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,64,balanced,0.5573333501815796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,64,balanced,0.3479199806849162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,64,power_law_1.2,0.5605567932128906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,64,balanced,0.37105600039164227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,64,balanced,0.2752853234608968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,64,balanced,0.40613865852355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,64,power_law_1.2,0.28922879695892334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,64,balanced,0.3397759993871053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,64,power_law_1.2,0.7339712142944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,64,balanced,0.47511998812357586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,64,power_law_1.2,0.3686271905899048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,64,balanced,0.599722663561503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,64,balanced,0.4389333327611287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,64,power_law_1.2,0.5602496147155762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,64,balanced,0.8490986824035645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,64,balanced,0.44680531819661456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,64,power_law_1.01,0.12654720544815062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,64,balanced,0.45610666275024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,64,balanced,1.0988372961680095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,64,balanced,0.4652959903081258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,64,power_law_1.01,0.1257151961326599
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,64,balanced,0.47996799151102704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,64,balanced,1.6091413497924805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,64,balanced,0.4972960154215495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,64,balanced,0.5914346774419149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,64,power_law_1.01,0.12602239847183228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,64,balanced,0.518234650293986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,64,power_law_1.01,0.21068799495697021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,64,balanced,0.630677342414856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,64,power_law_1.2,0.7252031803131104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,64,balanced,0.5512640078862509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,64,balanced,0.4413599967956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,64,power_law_1.01,0.28624639511108396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,64,balanced,2.121946652730306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,64,power_law_1.01,0.30558080673217775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,64,balanced,0.5864533185958862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,64,balanced,0.4819306532541911
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,64,power_law_1.01,0.35650560855865476
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,64,balanced,0.623520016670227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,64,power_law_1.2,1.0521408081054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,64,power_law_1.01,0.10474879741668701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,64,power_law_1.01,0.42676482200622556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,64,balanced,0.6927733421325684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,64,power_law_1.01,0.5033984184265137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,64,power_law_1.01,0.4352255821228027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,64,power_law_1.01,0.10489599704742432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,64,power_law_1.2,1.384224033355713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,64,power_law_1.01,0.4429056167602539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,64,power_law_1.01,0.10556800365447998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,64,power_law_1.01,0.45204482078552244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,64,power_law_1.01,0.10451200008392333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,64,balanced,0.702735980351766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,64,power_law_1.01,0.4597184181213379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,64,balanced,0.8515360355377197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,64,balanced,0.5546613136927286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,64,power_law_1.01,0.30429439544677733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,64,power_law_1.01,0.47800321578979493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,64,power_law_1.2,2.122796821594238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,64,balanced,0.9903786977132162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,64,power_law_1.01,0.42772479057312013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,64,power_law_1.01,0.4925695896148682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,64,power_law_1.01,0.6656320095062256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,64,balanced,1.3359252611796062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,64,power_law_1.01,0.43777918815612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,64,power_law_1.01,0.5157375812530518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,64,power_law_1.01,0.9684543609619141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,64,balanced,1.6190080642700195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,64,power_law_1.01,0.4454400062561035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,64,power_law_1.01,0.5487743854522705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,64,power_law_1.01,0.45631999969482423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,64,power_law_1.2,1.0262463569641114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,64,power_law_1.01,1.2486080169677733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,64,balanced,2.2300853729248047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,64,balanced,0.6985387007395426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,64,balanced,0.8328213691711426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,64,balanced,0.8429973125457764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,64,power_law_1.01,1.8945791244506835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,64,balanced,2.7990026473999023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,64,balanced,0.9740533034006754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,64,balanced,1.1633226871490479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,64,balanced,1.3326666355133057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,64,balanced,1.4488479296366374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,64,power_law_1.01,2.4237632751464844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,64,power_law_1.01,0.5920576095581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,64,balanced,4.073402722676595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,64,balanced,2.0562186241149902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,64,power_law_1.01,0.4652095794677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,64,power_law_1.01,0.6286015987396241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,64,power_law_1.01,0.10621440410614014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,64,power_law_1.01,0.4813695907592773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,64,power_law_1.01,0.7181183815002441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,64,balanced,2.6166559855143228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,64,power_law_1.01,0.1089792013168335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,64,power_law_1.01,0.5000319957733155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,64,power_law_1.01,0.9035200119018555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,64,power_law_1.01,0.11092480421066284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,64,balanced,5.3134613037109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,64,power_law_1.01,0.5220160007476806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,64,balanced,1.6078294118245442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,64,power_law_1.01,0.1388159990310669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,64,power_law_1.01,0.5590720176696777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,64,balanced,3.819098790486654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,64,power_law_1.2,1.4219200134277343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,64,power_law_1.01,0.6034560203552246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,64,balanced,2.2217706044514975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,64,power_law_1.01,0.6555263996124268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,64,power_law_1.01,0.7310463905334472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,64,balanced,2.8115787506103516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,64,balanced,4.982618649800618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,64,balanced,7.857615788777669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,64,power_law_1.01,0.9041024208068847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,64,power_law_1.01,1.078816032409668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,64,power_law_1.01,0.1999935984611511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,64,balanced,4.065269470214844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,64,power_law_1.01,1.4026880264282227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,64,power_law_1.01,0.27166080474853516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,64,power_law_1.01,0.2735424041748047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,64,power_law_1.01,1.7650047302246095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,64,balanced,7.366981506347656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,64,power_law_1.01,0.28664960861206057
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,64,balanced,5.2653811772664385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,64,power_law_1.01,0.2981247901916504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,64,power_law_1.01,1.0809856414794923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,64,power_law_1.01,2.410259246826172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,64,power_law_1.01,0.3089344024658203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,64,power_law_1.01,0.3238080024719238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,64,power_law_1.01,1.4197952270507812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,64,power_law_1.01,0.3462080001831055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,64,power_law_1.2,2.713657569885254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,64,power_law_1.01,3.052275276184082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,64,power_law_1.01,1.7515392303466797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,64,power_law_1.2,0.12760319709777831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,64,balanced,7.769999821980794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,64,balanced,10.324949264526367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,64,power_law_1.2,0.12758400440216064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,64,power_law_1.2,0.12654080390930175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,64,power_law_1.01,2.431782341003418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,64,power_law_1.2,0.10458240509033204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,64,power_law_1.2,2.081715202331543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,64,power_law_1.2,0.127948796749115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,64,power_law_1.01,4.362419128417969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,64,power_law_1.2,0.30410239696502683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,64,power_law_1.01,0.37518720626831054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,64,balanced,9.701583862304688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,64,power_law_1.2,2.7902528762817385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,64,power_law_1.01,3.0734527587890623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,128,balanced,0.06383466720581055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,128,balanced,0.06201066573460897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,128,balanced,0.1281760036945343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,128,balanced,0.1397119959195455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,128,balanced,0.17433599630991617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,64,balanced,10.189840316772461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,128,balanced,0.268394668896993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,128,balanced,0.04996266464392344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,128,balanced,0.2717439929644267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,128,balanced,0.05407466491063436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,128,balanced,0.08260266482830048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,128,balanced,0.09843732913335164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,128,balanced,0.13792533675829569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,128,balanced,0.2568053404490153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,128,balanced,0.2608426610628764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,128,balanced,0.26314133405685425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,128,balanced,0.26828267176946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,64,power_law_1.01,4.480339050292969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,128,balanced,0.27321066459019977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,128,balanced,0.28275734186172485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,128,balanced,0.2898186643918355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,64,power_law_1.2,0.42574081420898435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,128,balanced,0.29759466648101807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,128,balanced,0.31319467226664227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,64,power_law_1.2,0.10496640205383301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,128,balanced,0.3304106593132019
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,64,power_law_1.2,0.4367936134338379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,128,balanced,0.3478399912516276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,128,balanced,0.3794613281885783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,128,balanced,0.27529066801071167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,128,balanced,0.27962666749954224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,64,power_law_1.2,0.10472960472106933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,128,balanced,0.2821066578229268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,64,power_law_1.01,5.837497711181641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,128,balanced,0.2892799973487854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,128,balanced,0.2967039942741394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,64,power_law_1.2,0.10569599866867066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,128,balanced,0.30268800258636475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,128,balanced,0.3126666744550069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,64,power_law_1.01,5.952447891235352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,64,power_law_1.2,0.10414079427719117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,64,power_law_1.2,0.29719679355621337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,64,power_law_1.2,0.10874240398406983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,128,balanced,0.44707731405893963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,64,power_law_1.2,0.42513279914855956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,128,balanced,0.5139840046564738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,64,power_law_1.2,0.11002240180969239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,64,power_law_1.2,0.43525118827819825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,64,power_law_1.2,0.13987200260162352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,128,balanced,0.3288586735725403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,64,power_law_1.2,0.44330878257751466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,64,power_law_1.01,0.40523519515991213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,64,power_law_1.01,8.42078094482422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,64,power_law_1.2,0.4513023853302002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,64,power_law_1.01,0.44634242057800294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,64,power_law_1.2,0.4601344108581543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,64,power_law_1.01,8.686656188964843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,64,power_law_1.2,0.48026242256164553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,64,power_law_1.2,0.4971776008605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,64,power_law_1.2,0.5203904151916504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,128,balanced,0.3433440128962199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,128,balanced,0.3766133387883504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,64,power_law_1.2,0.19082239866256714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,128,balanced,0.43724266688028973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,128,balanced,0.4987786610921224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,64,power_law_1.01,0.4875840187072754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,64,power_law_1.2,0.5630335807800293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,128,balanced,0.6204906702041626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,64,power_law_1.2,0.4437248229980469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,128,balanced,0.7441813151041666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,64,power_law_1.01,0.5698624134063721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,64,power_law_1.2,0.6084928035736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,64,power_law_1.2,0.4520895957946777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,128,balanced,0.642959992090861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,64,power_law_1.01,0.7272192001342773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,64,power_law_1.2,0.4617023944854736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,64,power_law_1.2,0.6575424194335937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,64,power_law_1.2,0.4773695945739746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,64,power_law_1.01,0.8909440040588379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,64,power_law_1.01,12.043321228027343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,64,power_law_1.2,0.7492415904998779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,128,balanced,0.07261866827805837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,128,balanced,0.0718453327814738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,128,balanced,0.0788320004940033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,64,power_law_1.2,0.27146880626678466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,128,balanced,0.07743466893831889
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,64,power_law_1.01,11.37996826171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,64,power_law_1.2,0.9229567527770997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,64,power_law_1.2,0.2791872024536133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,128,balanced,0.99290664990743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,64,power_law_1.2,0.2888832092285156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,128,balanced,0.772810697555542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,128,balanced,1.2364426453908284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,64,power_law_1.2,0.29682559967041017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,128,balanced,1.0335520108540852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,64,power_law_1.2,0.3079999923706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,64,power_law_1.2,0.49409918785095214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,64,power_law_1.2,0.32968959808349607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,64,power_law_1.01,1.2109888076782227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,128,balanced,0.09949333469072978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,64,power_law_1.2,0.5166016101837159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,128,balanced,0.15244799852371216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,64,power_law_1.2,0.3496448040008545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,128,balanced,0.1564640005429586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,128,balanced,0.16025066375732422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,128,balanced,0.16368533174196878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,64,power_law_1.2,0.5557695865631104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,64,power_law_1.2,0.3689152002334595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,64,power_law_1.01,1.5405759811401367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,64,power_law_1.2,1.101036834716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,64,power_law_1.2,0.4115839958190918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,64,power_law_1.2,1.4382080078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,64,power_law_1.2,0.45296640396118165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,128,balanced,1.727834701538086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,128,balanced,1.2903093496958415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,64,power_law_1.01,2.1827264785766602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,64,power_law_1.2,0.494374418258667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,128,balanced,2.223200003306071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,128,balanced,1.8135414123535156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,64,power_law_1.2,1.8246784210205078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,64,power_law_1.2,0.5830719947814942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,128,balanced,0.16884799798329672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,128,balanced,0.17478932936986288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,128,balanced,0.1827626625696818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,64,power_law_1.01,2.8078399658203126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,128,balanced,2.339850743611654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,64,power_law_1.2,0.6010047912597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,128,balanced,3.260197321573893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,64,power_law_1.2,2.523846435546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,64,power_law_1.2,0.6326528072357178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,128,balanced,0.02754133443037669
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,128,balanced,0.027466667195161183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,128,balanced,0.028815999627113342
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,128,balanced,0.029578665892283123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,128,balanced,3.404447873433431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,128,balanced,0.033573334415753685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,64,power_law_1.2,0.7155712127685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,128,balanced,0.03605866680542628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,128,balanced,0.08282133440176646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,128,balanced,4.362298647562663
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,64,power_law_1.01,4.130393600463867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,64,power_law_1.2,0.941100788116455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,64,power_law_1.2,3.3611072540283202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,128,balanced,0.19131733973821005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,64,power_law_1.2,0.7528704166412353
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,128,balanced,0.20544532934824625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,64,power_law_1.2,1.0969087600708007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,128,balanced,0.2220053275426229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,128,balanced,0.23602133989334106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,128,balanced,0.2675039966901143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,64,power_law_1.2,0.9234751701354981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,128,balanced,0.3305973410606384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,128,balanced,0.3921440045038859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,128,balanced,0.5107146501541138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,128,balanced,0.052704001466433205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,128,balanced,0.6312160094579061
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,128,balanced,0.08860266208648682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,64,power_law_1.2,4.798271942138672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,128,balanced,0.08712533116340637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,64,power_law_1.01,5.41923828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,128,balanced,4.478309313456218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,128,power_law_1.01,0.07701119780540466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,128,power_law_1.01,0.07697920203208923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,64,power_law_1.2,1.4337920188903808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,128,power_law_1.01,0.1087231993675232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,64,power_law_1.2,1.2569472312927246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,128,power_law_1.01,0.12163840532302857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,128,power_law_1.01,0.08705919981002808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,64,power_law_1.2,1.7583488464355468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,64,power_law_1.2,6.260294342041016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,128,power_law_1.01,0.08725759983062745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,64,power_law_1.2,1.5838912010192872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,128,power_law_1.01,0.1504256010055542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,128,power_law_1.01,0.15639040470123292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,64,power_law_1.2,2.5323968887329102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,128,power_law_1.01,0.19159040451049805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,64,power_law_1.2,2.2263423919677736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,128,power_law_1.01,0.2655103921890259
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,128,balanced,0.06983999907970428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,128,power_law_1.01,0.2699903964996338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,128,balanced,0.054192001620928444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,128,power_law_1.01,0.15889919996261598
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,128,balanced,0.06028800209363302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,64,power_law_1.2,3.2459583282470703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,128,balanced,0.056421334544817604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,128,power_law_1.01,0.27279999256134035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,128,balanced,0.05215999980767568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,128,power_law_1.01,0.25505919456481935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,128,balanced,0.05342933535575867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,64,power_law_1.2,2.9477312088012697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,128,power_law_1.01,0.2781183958053589
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,128,balanced,0.05823466678460439
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,128,balanced,0.06664533416430156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,128,power_law_1.01,0.25949440002441404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,128,balanced,0.07467733323574066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,128,power_law_1.01,0.2807424068450928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,128,balanced,0.0992746651172638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,128,power_law_1.01,0.2632256031036377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,128,balanced,0.12044266859690349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,128,power_law_1.01,0.28751358985900877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,128,balanced,0.15915733575820923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,64,power_law_1.2,9.717926025390625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,128,balanced,0.20362667242685953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,128,power_law_1.01,0.2939647912979126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,128,balanced,0.2951520085334778
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,128,balanced,0.8776266574859619
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,128,power_law_1.01,0.3001983880996704
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,128,balanced,0.37823466459910077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,64,power_law_1.2,4.259609603881836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,128,power_law_1.01,0.3122175931930542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,128,balanced,1.112709363301595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,128,power_law_1.01,0.3263808012008667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,128,balanced,1.598090648651123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,128,power_law_1.01,0.342630410194397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,128,power_law_1.01,0.3771456003189087
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,128,balanced,2.082047939300537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,128,power_law_1.01,0.2670144081115723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,128,power_law_1.01,0.44303359985351565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,128,power_law_1.01,0.27238399982452394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,64,power_law_1.2,4.716831970214844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,128,power_law_1.01,0.5092991828918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,128,power_law_1.01,0.2800640106201172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,128,balanced,3.0533065795898438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,128,power_law_1.01,0.6457215785980225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,128,power_law_1.01,0.28906879425048826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,64,power_law_1.01,7.942918395996093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,128,power_law_1.01,0.2967103958129883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,128,power_law_1.01,0.8114751815795899
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,128,power_law_1.01,0.31132800579071046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,128,balanced,4.014143943786621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,128,power_law_1.01,1.1323328018188477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,128,power_law_1.01,0.3292736053466797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,64,power_law_1.2,13.232351684570313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,128,power_law_1.01,0.3466304063796997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,64,power_law_1.2,6.200569534301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,128,power_law_1.01,1.4448575973510742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,128,power_law_1.01,0.3821376085281372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,128,power_law_1.01,0.45252480506896975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,128,power_law_1.01,2.011155128479004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,128,power_law_1.01,0.07861120104789734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,128,balanced,0.5381120045979818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,128,power_law_1.01,2.6172224044799806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,128,power_law_1.01,0.08058879971504211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,128,balanced,0.7012800375620524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,128,power_law_1.01,0.08805760145187377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,128,power_law_1.2,0.0882752001285553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,128,power_law_1.01,0.5298240184783936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,128,power_law_1.2,0.08997759819030762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,128,power_law_1.2,0.15007359981536866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,128,power_law_1.01,0.6731967926025391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,128,power_law_1.01,3.8066497802734376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,64,power_law_1.2,9.555264282226563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,128,power_law_1.2,0.1535871982574463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,128,power_law_1.01,0.8344191551208496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,64,power_law_1.2,5.52929916381836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,128,power_law_1.2,0.19313280582427977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,64,power_law_1.01,10.60711669921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,128,power_law_1.2,0.2657279968261719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,128,power_law_1.01,1.1591808319091796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,128,power_law_1.01,0.08790400028228759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,128,power_law_1.2,0.07429119944572449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,128,power_law_1.2,0.27073919773101807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,128,power_law_1.01,0.10611200332641602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,128,power_law_1.2,0.07459200024604798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,128,power_law_1.01,1.4784192085266112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,128,power_law_1.01,0.14986239671707152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,128,power_law_1.2,0.11151360273361206
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,128,power_law_1.01,0.02842240035533905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,128,power_law_1.01,0.1552575945854187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,128,power_law_1.01,0.029068800806999206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,128,power_law_1.01,0.1575808048248291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,128,power_law_1.01,0.029631999135017396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,128,power_law_1.01,0.16170239448547363
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,128,power_law_1.01,0.030982398986816408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,128,power_law_1.01,0.16579840183258057
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,128,power_law_1.01,0.03387520015239716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,128,power_law_1.01,0.17367039918899535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,64,power_law_1.2,8.761529541015625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,128,power_law_1.01,0.036032000184059144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,128,power_law_1.01,0.18103679418563842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,128,power_law_1.01,5.036966323852539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,128,power_law_1.01,0.1883072018623352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,128,power_law_1.2,0.08094720244407654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,128,power_law_1.2,0.11716480255126953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,128,power_law_1.01,0.20381441116333007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,128,power_law_1.2,0.08352640271186829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,128,power_law_1.2,0.15249919891357422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,128,power_law_1.01,0.21986560821533202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,128,power_law_1.01,2.1036991119384765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,128,power_law_1.2,0.09057279825210571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,64,power_law_1.2,13.23055419921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,128,power_law_1.2,0.2543999910354614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,128,power_law_1.2,0.08763520121574402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,128,power_law_1.2,0.2594048023223877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,128,power_law_1.2,0.10784640312194824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,128,power_law_1.2,0.2730815887451172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,128,power_law_1.2,0.2624255895614624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,128,power_law_1.2,0.1506816029548645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,128,power_law_1.2,0.27715840339660647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,128,power_law_1.2,0.26689920425415037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,128,power_law_1.2,0.15523200035095214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,128,power_law_1.2,0.28126718997955324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,128,power_law_1.2,0.2709696054458618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,128,power_law_1.2,0.15802240371704102
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,128,power_law_1.2,0.28726398944854736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,128,power_law_1.2,0.27910399436950684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,128,power_law_1.2,0.16289279460906983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,128,power_law_1.2,0.2950527906417847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,128,power_law_1.01,0.23834240436553955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,128,power_law_1.2,0.2878528118133545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,128,power_law_1.2,0.301580810546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,128,power_law_1.01,0.2734976053237915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,128,power_law_1.2,0.2951551914215088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,128,power_law_1.2,0.31393918991088865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,128,power_law_1.01,0.3429631948471069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,128,power_law_1.2,0.3103679895401001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,128,power_law_1.01,2.7135744094848633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,128,power_law_1.2,0.3308480024337769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,128,power_law_1.2,0.3287103891372681
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,128,power_law_1.01,0.07817599773406983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,128,power_law_1.2,0.3453824043273926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,128,power_law_1.2,0.346233606338501
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,128,power_law_1.01,0.04953599870204926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,128,power_law_1.2,0.3802815914154053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,128,power_law_1.2,0.38073599338531494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,128,power_law_1.2,0.4465343952178955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,128,power_law_1.2,0.16631679534912108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,128,power_law_1.2,0.5167424201965332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,128,power_law_1.2,0.1747712016105652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,128,power_law_1.2,0.18170880079269408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,128,power_law_1.2,0.7037951946258545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,128,power_law_1.2,0.18907519578933715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,128,power_law_1.01,0.4133120059967041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,64,power_law_1.2,11.095308685302735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,128,power_law_1.2,0.20506880283355713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,128,power_law_1.2,0.8266048431396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,128,balanced,0.04785066843032837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,128,balanced,0.04665599763393402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,128,power_law_1.01,0.5500480175018311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,128,power_law_1.2,0.22645120620727538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,128,balanced,0.10057600339253743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,128,balanced,0.10486933588981628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,128,power_law_1.01,0.07823359966278076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,128,power_law_1.2,0.24149758815765382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,128,power_law_1.2,1.1759103775024413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,128,power_law_1.01,0.6795519828796387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,128,power_law_1.01,0.07580159902572632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,128,power_law_1.2,0.2752255916595459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,128,power_law_1.01,4.075340652465821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,128,power_law_1.2,0.35050880908966064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,128,power_law_1.01,0.9464192390441895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,128,power_law_1.2,1.5238271713256837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,128,power_law_1.2,0.4339968204498291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,128,power_law_1.01,1.2308992385864257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,128,power_law_1.2,0.5666175842285156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,128,power_law_1.2,2.3334335327148437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,128,balanced,0.1339466671148936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,128,power_law_1.2,0.7187200069427491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,128,balanced,0.23627199729283652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,128,power_law_1.01,1.7408767700195313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,128,balanced,0.2402720053990682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,128,power_law_1.2,0.4538432121276855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,128,balanced,0.24037333329518637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,128,balanced,0.2445440093676249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,128,power_law_1.01,5.268742370605469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,128,power_law_1.01,0.06373119950294495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,128,balanced,0.047925333182017006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,128,balanced,0.04791999856630961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,128,power_law_1.01,0.049465599656105044
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,128,power_law_1.01,0.051635199785232545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,128,power_law_1.01,0.051801598072052
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,128,power_law_1.2,0.02871679961681366
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,128,power_law_1.2,0.029407998919486998
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,128,power_law_1.2,0.030092799663543703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,128,power_law_1.2,1.0084416389465332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,128,power_law_1.2,0.03129599988460541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,128,power_law_1.2,2.8757503509521483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,128,balanced,0.07494399944941203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,128,balanced,0.0893440047899882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,128,power_law_1.2,0.034272000193595886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,128,balanced,0.12800000111262003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,128,power_law_1.01,2.321676826477051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,128,power_law_1.2,1.3179519653320313
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,128,power_law_1.2,0.036652800440788266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,128,power_law_1.01,0.05116159915924072
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,128,power_law_1.2,0.08005759716033936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,128,power_law_1.01,0.053523200750350955
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,128,power_law_1.2,0.05013120174407959
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,128,power_law_1.01,0.05708799958229065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,128,power_law_1.2,0.07960320115089417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,128,power_law_1.01,3.464985656738281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,128,power_law_1.01,0.06800640225410462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,128,power_law_1.2,0.5234496116638183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,128,power_law_1.2,0.07802240252494812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,128,power_law_1.01,0.0818943977355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,128,power_law_1.2,0.062080001831054686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,128,power_law_1.2,0.6984576225280762
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,128,power_law_1.01,0.10965119600296021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,128,power_law_1.2,0.050393599271774295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,128,power_law_1.01,0.13242239952087403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,128,power_law_1.2,0.05130879878997803
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,128,power_law_1.01,0.18029439449310303
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,128,power_law_1.2,0.05122560262680054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,128,power_law_1.2,1.871980857849121
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,128,power_law_1.2,0.05237759947776795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,128,power_law_1.2,4.169222259521485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,128,balanced,0.24862400690714517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,128,balanced,0.25682133436203003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,128,balanced,0.26290132602055866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,128,balanced,0.27102933327356976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,128,balanced,0.28335465987523395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,128,balanced,0.23434666792551676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,128,balanced,0.29765866200129193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,128,balanced,0.23646400372187296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,128,balanced,0.3141706585884094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,128,balanced,0.24035199483235678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,128,balanced,0.3428479830423991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,128,power_law_1.2,0.8456704139709472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,128,power_law_1.01,4.525811386108399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,128,power_law_1.2,6.019059371948242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,128,power_law_1.2,1.1986559867858886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,128,balanced,0.06637333333492279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,128,balanced,0.06916266679763794
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,128,balanced,0.07354666789372762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,128,balanced,0.07197866837183635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,128,power_law_1.2,2.562163162231445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,128,balanced,0.08932266632715861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,128,balanced,0.13878933588663736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,128,power_law_1.2,1.5327232360839844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,128,balanced,0.14451199769973755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,128,balanced,0.14630400141080221
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,128,balanced,0.24489066998163858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,128,balanced,0.4003359874089559
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,128,balanced,0.4577920039494832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,128,balanced,0.5724426507949829
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,128,power_law_1.01,0.23126399517059326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,128,power_law_1.2,2.305574417114258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,128,balanced,0.6844159762064616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,128,power_law_1.2,0.055232000350952146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,128,power_law_1.2,3.934540939331055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,128,balanced,0.9129760265350342
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,128,power_law_1.2,0.06042879819869995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,128,balanced,1.140437364578247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,128,power_law_1.2,2.9572479248046877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,128,balanced,0.14800533652305603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,128,balanced,1.5911839803059895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,128,balanced,0.15267200271288553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,128,balanced,0.24673600991566977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,128,balanced,0.15999466180801392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,128,balanced,0.25733866294225055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,128,balanced,0.16988267501195273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,128,balanced,0.2651413281758626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,128,balanced,0.17493333419164023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,128,balanced,0.271013339360555
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,128,balanced,2.0480213165283203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,128,balanced,0.28748265902201336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,128,balanced,0.301962673664093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,128,balanced,0.3182826638221741
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,128,power_law_1.01,0.3304192066192627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,128,power_law_1.2,4.258272171020508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,128,balanced,3.0036586125691733
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,128,power_law_1.01,0.431660795211792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,128,balanced,0.027061333258946735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,128,power_law_1.2,5.253433609008789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,128,power_law_1.01,0.6166783809661865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,128,power_law_1.01,0.07856000065803528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,128,balanced,0.1879733403523763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,128,balanced,3.9639838536580405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,128,balanced,0.205567995707194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,128,power_law_1.01,0.8337216377258301
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,128,power_law_1.01,0.05127679705619812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,128,power_law_1.01,0.07154560089111328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,128,power_law_1.01,0.05223039984703064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,128,power_law_1.01,0.07483519911766053
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,128,power_law_1.2,0.07272959947586059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,128,power_law_1.01,0.08908159732818603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,128,power_law_1.01,0.08218240141868591
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,128,power_law_1.01,0.09854080080986023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,128,balanced,0.027349332968393963
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,128,balanced,0.027434666951497395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,128,power_law_1.01,0.0815999984741211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,128,balanced,0.029535998900731403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,128,balanced,0.03164266546567281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,128,balanced,0.03573866685231527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,128,power_law_1.01,0.09565439820289612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,128,power_law_1.01,0.07852799892425537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,128,balanced,0.219376007715861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,128,power_law_1.01,0.1370303988456726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,128,balanced,0.24629332621892294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,128,power_law_1.01,0.13470720052719115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,128,balanced,0.3060426712036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,128,balanced,0.34693864981333417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,128,power_law_1.01,0.1407039999961853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,128,balanced,0.3631733258565267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,128,balanced,0.40991465250651044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,128,balanced,0.4778933525085449
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,128,balanced,0.47065067291259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,128,power_law_1.2,6.067782211303711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,128,balanced,0.5901546478271484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,128,balanced,0.5875466664632162
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,128,power_law_1.2,0.08776959776878357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,128,balanced,0.8225013415018717
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,128,power_law_1.2,0.11484800577163697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,128,power_law_1.01,0.13546240329742432
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,128,balanced,1.0423253377278645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,128,balanced,0.04797333478927612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,128,balanced,0.0773119976123174
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,128,balanced,0.07869333525498708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,128,balanced,0.08003733555475871
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,128,balanced,0.06446933249632518
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,128,balanced,0.05087999999523163
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,128,balanced,0.05862399935722351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,128,power_law_1.01,0.13742079734802246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,128,balanced,0.05275199810663859
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,128,balanced,0.054192001620928444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,128,balanced,0.050069332122802734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,128,power_law_1.01,0.16926079988479614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,128,balanced,0.05306133131186167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,128,balanced,0.05929600199063619
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,128,balanced,0.0666186660528183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,128,power_law_1.01,0.23308799266815186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,128,balanced,0.08482666810353597
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,128,balanced,0.10774933298428853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,128,power_law_1.01,0.23780479431152343
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,128,balanced,0.14285332957903543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,128,balanced,0.17999466260274252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,128,power_law_1.2,0.13680000305175782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,128,power_law_1.01,0.2399616003036499
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,128,balanced,0.262992004553477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,128,power_law_1.2,0.19363839626312257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,128,power_law_1.01,0.24420480728149413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,128,power_law_1.01,0.24679040908813477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,128,power_law_1.2,0.2420351982116699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,128,power_law_1.01,0.2557055950164795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,128,power_law_1.01,0.14412800073623658
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,128,power_law_1.2,0.35649919509887695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,128,power_law_1.01,0.2620863914489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,128,power_law_1.01,0.14632960557937622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,128,balanced,0.7067253589630127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,128,power_law_1.01,0.2691456079483032
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,128,power_law_1.01,0.15100799798965453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,128,power_law_1.01,0.28255999088287354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,128,power_law_1.01,0.15945600271224974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,128,power_law_1.01,0.29692161083221436
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,128,power_law_1.01,0.02778880000114441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,128,power_law_1.01,0.1673472046852112
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,128,balanced,0.33673067887624103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,128,power_law_1.01,0.3136768102645874
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,128,power_law_1.01,0.028089600801467895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,128,power_law_1.01,0.17351679801940917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,128,power_law_1.01,0.34552319049835206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,128,power_law_1.01,0.1867583990097046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,128,power_law_1.01,0.4052608013153076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,128,power_law_1.01,0.203385591506958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,128,power_law_1.01,0.46744961738586427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,128,power_law_1.01,0.2164992094039917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,128,power_law_1.01,0.25191679000854494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,128,power_law_1.01,0.5868864059448242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,128,power_law_1.01,0.31909120082855225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,128,power_law_1.01,0.7376128196716308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,128,power_law_1.01,0.38725759983062746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,128,balanced,0.4758506615956624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,128,power_law_1.01,0.029023998975753786
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,128,power_law_1.01,0.030502399802207945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,128,power_law_1.2,0.47207040786743165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,128,power_law_1.01,0.03343999981880188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,128,balanced,0.9428906440734863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,128,power_law_1.01,0.035308799147605895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,128,power_law_1.2,0.6945151805877685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,128,power_law_1.01,0.2317568063735962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,128,balanced,1.17575470606486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,128,power_law_1.01,0.0463808000087738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,128,balanced,1.4968959490458171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,128,power_law_1.01,0.23582079410552978
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,128,power_law_1.01,0.07428479790687562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,128,power_law_1.2,0.9389375686645508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,128,balanced,1.6492533683776855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,128,power_law_1.01,0.5201216220855713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,128,balanced,1.9500853220621746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,128,power_law_1.01,0.07278720140457154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,128,power_law_1.2,0.08069760203361512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,128,power_law_1.01,0.6328447818756103
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,128,power_law_1.01,0.07360640168190002
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,128,balanced,2.1214613914489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,128,power_law_1.01,0.05920000076293945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,128,power_law_1.01,0.8933440208435058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,128,balanced,2.8582452138264975
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,128,power_law_1.01,0.04769279956817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,128,power_law_1.01,0.049753600358963014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,128,balanced,3.086912155151367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,128,power_law_1.01,1.1668352127075194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,128,power_law_1.01,0.049465599656105044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,128,power_law_1.01,1.0341440200805665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,128,balanced,3.7624692916870117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,128,power_law_1.01,1.6382976531982423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,128,power_law_1.2,0.07969279885292054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,128,balanced,4.058229446411133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,128,power_law_1.2,0.13358080387115479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,128,power_law_1.2,0.07226240038871765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,128,power_law_1.2,0.13411200046539307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,128,power_law_1.2,0.07486720085144043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,128,power_law_1.2,0.16885759830474853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,128,power_law_1.2,0.08108800053596496
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,128,power_law_1.2,0.232422399520874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,128,power_law_1.2,0.07820159792900086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,128,balanced,0.6202880144119263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,128,power_law_1.2,0.23644158840179444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,128,power_law_1.2,0.0961408019065857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,128,power_law_1.2,0.054502397775650024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,128,power_law_1.2,0.23900799751281737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,128,power_law_1.01,1.2962623596191407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,128,power_law_1.2,0.05603839755058289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,128,power_law_1.2,0.2434688091278076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,128,power_law_1.2,0.08994560241699219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,128,power_law_1.01,2.166592025756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,128,power_law_1.2,0.10029439926147461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,128,power_law_1.2,0.24567039012908937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,128,power_law_1.01,1.8687040328979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,128,power_law_1.2,0.13714560270309448
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,128,power_law_1.01,0.23868160247802733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,128,power_law_1.2,0.2309312105178833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,128,power_law_1.01,0.24277760982513427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,128,power_law_1.2,0.23503360748291016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,128,power_law_1.01,2.3868736267089843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,128,power_law_1.01,3.14138240814209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,128,power_law_1.01,0.24552319049835206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,128,power_law_1.2,0.2384768009185791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,128,power_law_1.01,0.25589120388031006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,128,power_law_1.2,0.2423232078552246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,128,power_law_1.01,0.26368639469146726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,128,power_law_1.2,0.24651520252227782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,128,power_law_1.01,0.27124478816986086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,128,power_law_1.2,0.254035210609436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,128,power_law_1.2,0.25470719337463377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,128,power_law_1.01,0.28492159843444825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,128,power_law_1.2,0.26099200248718263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,128,power_law_1.2,0.26375679969787597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,128,power_law_1.01,0.3006079912185669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,128,power_law_1.01,4.242713546752929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,128,power_law_1.2,0.2684927940368652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,128,power_law_1.2,0.269868803024292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,128,power_law_1.01,0.3159168004989624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,128,power_law_1.2,0.2822720050811768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,128,power_law_1.2,0.28463358879089357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,128,power_law_1.01,0.3476736068725586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,128,power_law_1.2,0.297875189781189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,128,power_law_1.2,0.3010560035705566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,128,power_law_1.2,0.3146048069000244
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,128,power_law_1.2,0.02573440074920654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,128,power_law_1.2,0.31687679290771487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,128,power_law_1.2,0.34456961154937743
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,128,power_law_1.2,0.026150399446487428
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,128,power_law_1.2,0.02705279886722565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,128,power_law_1.01,3.6361152648925783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,128,power_law_1.01,0.050483202934265135
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,128,power_law_1.2,0.028512001037597656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,128,power_law_1.01,0.049055999517440795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,128,balanced,0.027445333699385326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,128,balanced,0.03884266565243403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,128,balanced,0.03902933249870936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,128,power_law_1.01,0.052070397138595584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,128,balanced,0.043807998299598694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,128,power_law_1.2,0.13744640350341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,128,balanced,0.05595199763774872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,128,balanced,0.05590933561325073
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,128,power_law_1.01,0.06472960114479065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,128,power_law_1.2,0.14177279472351073
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,128,power_law_1.01,0.07501440048217774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,128,power_law_1.01,0.4078847885131836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,128,power_law_1.2,0.14454400539398193
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,128,power_law_1.01,0.09556480050086975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,128,power_law_1.01,0.47490558624267576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,128,power_law_1.01,4.719961547851563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,128,power_law_1.2,0.1487488031387329
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,128,power_law_1.01,0.11984000205993653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,128,power_law_1.2,0.4092671871185303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,128,power_law_1.2,0.15159039497375487
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,128,power_law_1.01,0.1614016056060791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,128,power_law_1.2,0.16030720472335816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,128,power_law_1.2,0.48035202026367185
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,128,power_law_1.01,0.20251519680023194
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,128,power_law_1.2,0.03112959861755371
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,128,power_law_1.2,0.16692479848861694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,128,power_law_1.2,0.6151360034942627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,128,power_law_1.01,0.2919487953186035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,128,balanced,0.05834133426348368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,128,power_law_1.2,0.17420159578323363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,128,balanced,0.060234665870666504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,128,balanced,0.06244266529877981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,128,balanced,0.06246933341026306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,128,power_law_1.2,0.18805760145187378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,128,balanced,0.06630933284759521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,128,power_law_1.01,0.3894399881362915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,128,power_law_1.2,0.8162495613098144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,128,power_law_1.2,0.20848000049591064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,128,power_law_1.2,0.3473599910736084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,128,power_law_1.01,0.545036792755127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,128,power_law_1.2,1.1016511917114258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,128,power_law_1.2,0.4145535945892334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,128,power_law_1.01,0.7198080062866211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,128,power_law_1.2,0.4845439910888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,128,balanced,0.03769599894682566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,128,power_law_1.2,1.3588607788085938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,128,power_law_1.2,0.6573823928833008
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,128,power_law_1.2,0.03377920091152191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,128,power_law_1.2,0.8016192436218261
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,128,power_law_1.2,0.04633600115776062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,128,power_law_1.2,2.0593215942382814
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,128,power_law_1.2,0.07367039918899536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,128,power_law_1.2,0.22081921100616456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,128,power_law_1.2,1.0831487655639649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,128,power_law_1.01,0.6070655822753906
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,128,power_law_1.2,0.07340800166130065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,128,power_law_1.2,0.25640320777893066
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,128,power_law_1.2,0.0737600028514862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,128,power_law_1.2,0.3247423887252808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,128,power_law_1.2,2.688915252685547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,128,power_law_1.2,1.409209632873535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,128,power_law_1.2,0.05939840078353882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,128,power_law_1.2,0.39486079216003417
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,128,power_law_1.2,0.048441600799560544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,128,power_law_1.2,0.5270527839660645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,128,power_law_1.2,0.054067200422286986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,128,balanced,0.06863999863465627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,128,power_law_1.2,1.9900863647460938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,128,balanced,0.07260799904664357
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,128,power_law_1.2,0.049107199907302855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,128,balanced,0.07897066573301952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,128,power_law_1.2,0.6530879974365235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,128,balanced,0.08478400111198425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,128,balanced,0.09099200367927551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,128,balanced,0.029338667790095013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,128,balanced,0.10327466328938802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,128,balanced,0.04012800008058548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,128,balanced,0.1267093320687612
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,128,balanced,0.041797334949175514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,128,power_law_1.2,4.012595367431641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,128,balanced,0.04385599990685781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,128,balanced,0.15101333459218344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,128,balanced,0.05816533168156942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,128,power_law_1.2,0.9430399894714355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,128,balanced,0.19419733683268228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,128,balanced,0.05982933441797892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,128,balanced,0.06204266846179962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,128,balanced,0.22456000248591104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,128,balanced,0.0640533318122228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,128,balanced,0.06440000236034393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,128,power_law_1.01,0.7637311935424804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,128,power_law_1.2,2.715776062011719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,128,balanced,0.040591999888420105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,128,balanced,0.04197333256403605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,128,balanced,0.042064001162846885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,128,power_law_1.2,1.1981823921203614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,128,balanced,0.046122665206591286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,128,balanced,0.04601066807905833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,128,balanced,0.04791999856630961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,128,balanced,0.04800533254941305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,128,power_law_1.01,1.0238975524902343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,128,balanced,0.05211733281612396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,128,balanced,0.05187733471393585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,128,balanced,0.055311997731526695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,128,balanced,0.056143999099731445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,128,balanced,0.06043733159701029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,128,power_law_1.2,1.7773439407348632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,128,power_law_1.01,1.3121024131774903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,128,power_law_1.2,5.636953735351563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,128,power_law_1.2,3.9895488739013674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,128,balanced,0.3096533417701721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,128,balanced,0.06613866488138835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,128,power_law_1.01,1.9231744766235352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,128,balanced,0.39344000816345215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,128,balanced,0.552512009938558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,128,balanced,0.7086613178253174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,128,balanced,1.0245013236999512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,128,power_law_1.01,2.483475112915039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,128,balanced,0.06678399940331776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,128,balanced,0.07189333438873291
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,128,power_law_1.2,0.050470399856567386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,128,balanced,0.07885866860548656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,128,balanced,1.340986728668213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,128,balanced,0.025248001019159954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,128,balanced,0.09086400270462036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,128,balanced,0.025386666258176167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,128,power_law_1.2,0.04891520142555237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,128,power_law_1.2,5.337958526611328
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,128,balanced,0.02534399926662445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,128,balanced,0.027466667195161183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,128,balanced,0.028213332096735638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,128,power_law_1.2,0.05322880148887634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,128,balanced,0.033557333052158356
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,128,balanced,0.035562666753927864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,128,balanced,0.03924266745646795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,128,power_law_1.2,2.3060415267944334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,128,balanced,0.07039999961853027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,128,balanced,0.07292266686757405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,128,balanced,0.07666666805744171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,128,balanced,0.08206399778525035
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,128,balanced,0.0890880028406779
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,128,balanced,0.09518399834632874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,128,balanced,0.1090666651725769
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,128,balanced,0.1318933367729187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,128,balanced,0.15586666266123453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,128,balanced,0.20065067211786905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,128,balanced,0.23830399910608926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,128,balanced,0.3370986779530843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,128,power_law_1.2,3.4959743499755858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,128,balanced,0.11356266339619954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,128,balanced,0.4243359963099162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,128,balanced,0.1376426617304484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,128,balanced,0.17212265729904175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,128,balanced,0.5963093439737955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,128,power_law_1.01,0.0296640008687973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,128,balanced,0.20780799786249796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,128,balanced,0.3002240061759949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,128,balanced,0.7703839937845866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,128,power_law_1.01,0.03857280015945434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,128,balanced,0.38199468453725177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,128,power_law_1.01,0.038553598523139956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,128,balanced,1.115781307220459
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,128,balanced,0.5418293476104736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,128,power_law_1.01,0.04408960044384003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,128,balanced,0.700709342956543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,128,power_law_1.01,0.05694720149040222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,128,balanced,1.460869312286377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,128,power_law_1.01,0.058284801244735715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,128,balanced,1.0190292994181316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,128,power_law_1.2,4.632998275756836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,128,power_law_1.01,0.060063999891281125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,128,power_law_1.01,0.03600000143051148
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,128,power_law_1.01,0.06162559986114502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,128,power_law_1.01,0.038252800703048706
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,128,balanced,1.3367466926574707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,128,power_law_1.01,0.026156800985336303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,128,power_law_1.01,0.06273279786109924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,128,power_law_1.01,0.040870401263237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2048,768,8,128,1,128,power_law_1.2,0.025183999538421632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,128,power_law_1.01,0.03734399974346161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,128,power_law_1.01,0.04026240110397339
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,128,power_law_1.2,0.06330879926681518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,128,power_law_1.01,0.037536001205444335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,128,power_law_1.01,0.04376960098743439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,128,power_law_1.01,0.04095999896526337
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,128,power_law_1.2,0.07639679908752442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,128,power_law_1.01,3.5626049041748047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,128,power_law_1.01,0.05399680137634277
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,128,power_law_1.2,0.09941759705543518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,128,power_law_1.01,0.0548799991607666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,128,power_law_1.01,0.0573248028755188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,128,power_law_1.2,0.12536959648132323
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,128,power_law_1.01,0.024140800535678863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,128,power_law_1.01,0.058284801244735715
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,128,power_law_1.2,0.17534079551696777
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,128,power_law_1.01,0.023904000222682954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,128,power_law_1.01,0.023999999463558196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,128,power_law_1.2,0.22525439262390137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,128,power_law_1.01,0.0640064001083374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2048,768,8,128,1,128,power_law_1.2,0.037088000774383546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,128,power_law_1.01,0.06691200137138367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,128,power_law_1.2,0.3130944013595581
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,128,power_law_1.01,4.67064323425293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2048,768,8,128,1,128,power_law_1.2,0.03710080087184906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,128,power_law_1.01,0.06999040246009827
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,128,power_law_1.01,0.04456959962844849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2048,768,8,128,1,128,power_law_1.2,0.040940800309181215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,128,power_law_1.01,0.07248640060424805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,128,power_law_1.01,0.047142401337623596
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,128,power_law_1.01,0.07831680178642272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,128,power_law_1.01,0.04684799909591675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,128,power_law_1.01,0.08292480111122132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,128,power_law_1.01,0.049721598625183105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,128,power_law_1.01,0.08899840116500854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,128,power_law_1.01,0.04955520033836365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,128,power_law_1.01,0.06039040088653565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,128,power_law_1.01,0.0992255985736847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,128,power_law_1.01,0.0523904025554657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,128,power_law_1.01,0.06259199976921082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,128,power_law_1.01,0.12011519670486451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,128,power_law_1.01,0.05649920105934143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,128,power_law_1.01,0.0648576021194458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,128,power_law_1.01,0.1447167992591858
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,128,power_law_1.01,0.05902720093727112
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,128,balanced,0.03660800059636434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,128,power_law_1.01,0.06833919882774353
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,128,balanced,0.03761066744724909
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,128,power_law_1.01,0.19041919708251953
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,128,balanced,0.03745066622893015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,128,power_law_1.01,0.07088639736175537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,128,power_law_1.01,0.23665919303894042
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,128,power_law_1.2,0.4164544105529785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,128,power_law_1.01,0.33564798831939696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2048,768,8,128,1,128,power_law_1.2,0.054451197385787964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2048,768,8,128,1,128,power_law_1.2,0.05492479801177978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,128,power_law_1.01,0.4331520080566406
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2048,768,8,128,1,128,power_law_1.2,0.05800319910049438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2048,768,8,128,1,128,power_law_1.2,0.058303999900817874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,128,power_law_1.01,0.6241727828979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2048,768,8,128,1,128,power_law_1.2,0.060684800148010254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2048,768,8,128,1,128,power_law_1.2,0.06166399717330932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,128,power_law_1.01,0.8146047592163086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2048,768,8,128,1,128,power_law_1.2,0.06480000019073487
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,128,power_law_1.01,0.0646336019039154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,128,balanced,0.037791999677817024
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,128,balanced,0.03766933331886927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,128,power_law_1.01,0.07218559980392455
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,128,balanced,0.03724266588687897
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,128,balanced,0.03753600021203359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,128,power_law_1.01,0.07651200294494628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,128,balanced,0.037952000896135964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,128,balanced,0.03755199909210205
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,128,balanced,0.04187199970086416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,128,power_law_1.01,0.08844159841537476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,128,balanced,0.04376000165939331
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,128,balanced,0.04814933240413666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,128,power_law_1.01,0.11292799711227416
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,128,balanced,0.05300800005594889
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,128,balanced,0.06316266457239787
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,128,power_law_1.01,0.13637759685516357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,128,power_law_1.01,0.17745920419692993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,128,power_law_1.01,0.21544320583343507
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,128,power_law_1.01,0.025894400477409363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,128,power_law_1.01,0.309555196762085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,128,power_law_1.01,0.027001601457595826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2048,768,8,128,1,128,power_law_1.2,0.06807680130004883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,128,power_law_1.01,0.07722880244255066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,128,power_law_1.01,1.210476779937744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,128,power_law_1.01,0.08300799727439881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,128,power_law_1.01,0.09013760089874268
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,128,power_law_1.2,0.6118080139160156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,128,power_law_1.01,0.10106879472732544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,128,power_law_1.01,1.5616576194763183
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,128,balanced,0.0729066679875056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,128,power_law_1.01,0.12353919744491577
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,128,balanced,0.0993280013402303
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,128,power_law_1.2,0.8137791633605957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,2048,768,8,128,1,128,power_law_1.2,0.03691520094871521
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,128,balanced,0.12954666217168173
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,128,balanced,0.18716800212860107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,2048,768,8,128,1,128,power_law_1.2,0.039238399267196654
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,128,balanced,0.24493332703908285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,128,balanced,0.035818666219711304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,2048,768,8,128,1,128,power_law_1.2,0.04007039964199066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,128,balanced,0.060122668743133545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,128,balanced,0.05765866736570994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,128,balanced,0.06941866874694824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,2048,768,8,128,1,128,power_law_1.2,0.04010879993438721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,128,balanced,0.09353599945704143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,128,balanced,0.09662933150927226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2048,768,8,128,1,128,power_law_1.2,0.02812800109386444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2048,768,8,128,1,128,power_law_1.2,0.07129600048065185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,128,balanced,0.09893866380055745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,128,power_law_1.01,0.39772160053253175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,128,balanced,0.09937600294748943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,128,balanced,0.10145599643389384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2048,768,8,128,1,128,power_law_1.2,0.03973119854927063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,128,balanced,0.10522133111953735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,128,balanced,0.10981866717338562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2048,768,8,128,1,128,power_law_1.2,0.03917439877986908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,128,power_law_1.01,0.5764544010162354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2048,768,8,128,1,128,power_law_1.2,0.044537600874900815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,128,power_law_1.01,0.7500160217285157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,2048,768,8,128,1,128,power_law_1.2,0.021612800657749176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,128,power_law_1.01,0.14410879611968994
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,2048,768,8,128,1,128,power_law_1.2,0.021971200406551362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,128,power_law_1.01,0.1852671980857849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,128,power_law_1.01,1.105900764465332
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,2048,768,8,128,1,128,power_law_1.2,0.021798400580883025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,128,power_law_1.01,0.22269439697265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,2048,768,8,128,1,128,power_law_1.2,0.023647999763488768
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,128,power_law_1.01,0.3156543970108032
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,128,power_law_1.01,0.03294720053672791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,2048,768,8,128,1,128,power_law_1.2,0.04431999921798706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,2048,768,8,128,1,128,power_law_1.2,0.025171199440956117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,128,power_law_1.01,0.03405439853668213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,128,power_law_1.01,1.4433728218078614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,128,power_law_1.01,0.3999232053756714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,2048,768,8,128,1,128,power_law_1.2,0.04490239918231964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,128,balanced,0.04171200096607208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,2048,768,8,128,1,128,power_law_1.2,0.031206399202346802
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,128,power_law_1.01,0.03526400029659271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,128,balanced,0.11345600088437398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,2048,768,8,128,1,128,power_law_1.2,0.04681600034236908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,128,balanced,0.12147200107574463
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,2048,768,8,128,1,128,power_law_1.2,0.03227519989013672
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,128,power_law_1.01,0.035648000240325925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,128,power_law_1.01,0.5878592014312745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,2048,768,8,128,1,128,power_law_1.2,0.04750719964504242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,2048,768,8,128,1,128,power_law_1.2,0.03361279964447021
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,128,power_law_1.01,0.03553279936313629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2048,768,8,128,1,128,power_law_1.2,0.057657599449157715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,2048,768,8,128,1,128,power_law_1.2,0.04924800097942352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,2048,768,8,128,1,128,power_law_1.2,0.03357439935207367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,128,power_law_1.01,0.755456018447876
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2048,768,8,128,1,128,power_law_1.2,0.05815039873123169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,2048,768,8,128,1,128,power_law_1.2,0.05025920271873474
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,2048,768,8,128,1,128,power_law_1.2,0.03423359990119934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2048,768,8,128,1,128,power_law_1.2,0.061286401748657224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,2048,768,8,128,1,128,power_law_1.2,0.052960002422332765
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,128,power_law_1.2,0.034892800450325015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2048,768,8,128,1,128,power_law_1.2,0.061868798732757566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,128,power_law_1.01,1.1221311569213868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,2048,768,8,128,1,128,power_law_1.2,0.05719040036201477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,128,power_law_1.2,0.03505280017852783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2048,768,8,128,1,128,power_law_1.2,0.06261119842529297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,2048,768,8,128,1,128,power_law_1.2,0.05862399935722351
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,128,power_law_1.2,0.03518719971179962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2048,768,8,128,1,128,power_law_1.2,0.06496000289916992
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,2048,768,8,128,1,128,power_law_1.2,0.06515200138092041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2048,768,8,128,1,128,power_law_1.2,0.07710080146789551
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,128,power_law_1.2,0.035724800825119016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2048,768,8,128,1,128,power_law_1.2,0.06673280000686646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,128,power_law_1.01,1.4780735969543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2048,768,8,128,1,128,power_law_1.2,0.08216959834098816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2048,768,8,128,1,128,power_law_1.2,0.0707647979259491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2048,768,8,128,1,128,power_law_1.2,0.08954240083694458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2048,768,8,128,1,128,power_law_1.2,0.07383040189743043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2048,768,8,128,1,128,power_law_1.2,0.10069119930267334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2048,768,8,128,1,128,power_law_1.2,0.07906559705734253
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,2048,768,8,128,1,128,power_law_1.01,0.03616639971733093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2048,768,8,128,1,128,power_law_1.2,0.0834879994392395
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,2048,768,8,128,1,128,power_law_1.01,0.035283198952674864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2048,768,8,128,1,128,power_law_1.2,0.08995199799537659
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,2048,768,8,128,1,128,power_law_1.01,0.03580160140991211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2048,768,8,128,1,128,power_law_1.2,0.10053119659423829
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,2048,768,8,128,1,128,power_law_1.01,0.035743999481201175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2048,768,8,128,1,128,power_law_1.2,0.1215999960899353
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,128,power_law_1.01,0.03633280098438263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2048,768,8,128,1,128,power_law_1.2,0.1455423951148987
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,128,power_law_1.01,0.03692800104618073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,2048,768,8,128,1,128,power_law_1.2,0.07034879922866821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,128,balanced,0.05780800183614095
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,128,power_law_1.01,0.0374208003282547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,128,balanced,0.05964266757170359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2048,768,8,128,1,128,power_law_1.2,0.19741439819335938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,128,balanced,0.13037866353988647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,2048,768,8,128,1,128,power_law_1.2,0.07779200077056884
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,128,power_law_1.01,0.04100480079650879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,128,balanced,0.14469866951306662
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,128,balanced,0.15703466534614563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,128,power_law_1.01,0.04271360039710999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,128,balanced,0.1812480092048645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2048,768,8,128,1,128,power_law_1.2,0.12250239849090576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,128,balanced,0.22116265694300333
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,128,power_law_1.01,0.047884801030159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,128,balanced,0.2658346692721049
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2048,768,8,128,1,128,power_law_1.2,0.14535679817199706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,128,power_law_1.01,0.05321599841117859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2048,768,8,128,1,128,power_law_1.2,0.18793599605560302
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,128,power_law_1.01,0.0681984007358551
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,128,power_law_1.01,0.08320000171661376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2048,768,8,128,1,128,power_law_1.2,0.23211519718170165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,128,power_law_1.01,0.11431679725646973
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,128,power_law_1.01,0.15040639638900757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,128,balanced,0.06655466556549072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,128,balanced,0.09514133135477702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,128,balanced,0.09715732932090759
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,128,power_law_1.01,0.21416959762573243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,2048,768,8,128,1,128,power_law_1.2,0.08858240246772767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,2048,768,8,128,1,128,power_law_1.2,0.11178879737854004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,128,power_law_1.01,0.28194561004638674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,2048,768,8,128,1,128,power_law_1.2,0.13664000034332274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,128,balanced,0.35884801546732586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,128,balanced,0.4477173487345378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,128,balanced,0.6185386578241984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,128,balanced,0.784506638844808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,128,balanced,1.1226080258687336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,128,balanced,0.10109866658846538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,128,balanced,1.4591253598531086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,128,balanced,0.10334933797518413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,128,balanced,0.10598933696746826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,128,balanced,0.10734400153160095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,128,balanced,0.11343466242154439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,128,balanced,0.11974400281906128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2048,768,8,128,1,128,power_law_1.2,0.3335423946380615
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,128,balanced,0.12567466497421265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,128,balanced,2.135770638783773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,2048,768,8,128,1,128,power_law_1.2,0.18222719430923462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,2048,768,8,128,1,128,power_law_1.2,0.2273535966873169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,128,balanced,2.8229811986287436
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,2048,768,8,128,1,128,power_law_1.2,0.035648000240325925
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,128,balanced,0.02733866622050603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,128,power_law_1.01,0.03457919955253601
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,2048,768,8,128,1,128,power_law_1.2,0.03552640080451965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2048,768,8,128,1,128,power_law_1.2,0.24333438873291016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2048,768,8,128,1,128,power_law_1.2,0.4312704086303711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,128,power_law_1.01,0.059308797121047974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,2048,768,8,128,1,128,power_law_1.2,0.3272576093673706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,2048,768,8,128,1,128,power_law_1.2,0.03668479919433594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,128,power_law_1.01,0.05752320289611816
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,2048,768,8,128,1,128,power_law_1.2,0.04012160003185272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,128,power_law_1.01,0.06819199919700622
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,2048,768,8,128,1,128,power_law_1.2,0.04219520092010498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,128,balanced,0.1357493301232656
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,128,power_law_1.01,0.09217919707298279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,128,balanced,0.14596800009409586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,2048,768,8,128,1,128,power_law_1.2,0.04826239943504333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,128,balanced,0.15863999724388123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,128,power_law_1.01,0.0935808002948761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,128,balanced,0.18235733111699423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,2048,768,8,128,1,128,power_law_1.2,0.05780479907989502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,128,balanced,0.22985599438349405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,128,power_law_1.01,0.09616640210151672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,128,balanced,0.2763040065765381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,128,power_law_1.01,0.09816960096359253
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,128,balanced,0.3752426703770955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2048,768,8,128,1,128,power_law_1.2,0.35796480178833007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,128,power_law_1.01,0.1019327998161316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,2048,768,8,128,1,128,power_law_1.2,0.07351040244102477
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,128,balanced,0.02739733209212621
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,128,balanced,0.027488000690937042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,128,power_law_1.01,0.10329600572586059
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,128,balanced,0.029285334050655365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,128,balanced,0.03086400032043457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,128,power_law_1.01,0.10782719850540161
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,128,balanced,0.037733333806196846
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,128,balanced,0.039605334401130676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,128,power_law_1.01,0.11412479877471923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2048,768,8,128,1,128,power_law_1.2,0.6279295921325684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,128,power_law_1.01,0.11937919855117798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,128,power_law_1.01,0.1308608055114746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,2048,768,8,128,1,128,power_law_1.2,0.09154559969902039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,128,power_law_1.01,0.1481727957725525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,128,power_law_1.01,0.1625856041908264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2048,768,8,128,1,128,power_law_1.2,0.4614975929260254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,128,power_law_1.01,0.18320640325546264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,128,balanced,0.4669813315073649
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,2048,768,8,128,1,128,power_law_1.2,0.41660161018371583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,128,balanced,0.6502293348312378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,128,balanced,0.8303253650665283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,128,balanced,0.0408746674656868
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,128,balanced,0.04735999802748362
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,128,balanced,0.04765866696834564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,128,balanced,1.184448003768921
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,128,balanced,1.5448266665140789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2048,768,8,128,1,128,power_law_1.2,0.6640448093414306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,2048,768,8,128,1,128,power_law_1.2,0.6076863765716553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,128,balanced,2.2622933387756348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,128,balanced,2.99071470896403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,128,power_law_1.01,0.03565439879894257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,128,balanced,0.05273599922657013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,128,power_law_1.01,0.056383997201919556
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,128,balanced,0.05503466725349426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,128,balanced,0.05795733133951823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,128,balanced,0.0588319996992747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,128,power_law_1.01,0.05594879984855652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,128,balanced,0.06788266698519389
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,128,balanced,0.06774400174617767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,128,balanced,0.07011733452479045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,128,power_law_1.01,0.06717439889907836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,128,power_law_1.01,0.22390398979187012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,128,power_law_1.01,0.09393919706344604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,128,power_law_1.01,0.27388160228729247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,128,power_law_1.01,0.09598079919815064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,128,power_law_1.01,0.3733504056930542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,128,balanced,0.0402453343073527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,128,balanced,0.04152533411979675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,128,power_law_1.01,0.48140802383422854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,128,balanced,0.041536000867684685
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,128,balanced,0.04182399809360504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,128,balanced,0.043893332282702126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,128,balanced,0.043706665436426796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,128,power_law_1.01,0.6829567909240722
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,128,balanced,0.04558399816354116
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,128,balanced,0.050186668833096824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,2048,768,8,128,1,128,power_law_1.2,0.7921408176422119
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,128,balanced,0.05417599777380625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,128,balanced,0.06352533400058746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,128,balanced,0.07171733180681865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,128,balanced,0.07207466661930084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,128,balanced,0.0969546635945638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,128,balanced,0.12583999832471213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,128,balanced,0.1817013422648112
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,128,balanced,0.23804267247517905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,128,power_law_1.01,0.09949439764022827
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,128,balanced,0.33716265360514325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,128,power_law_1.01,0.101254403591156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,128,balanced,0.4352426528930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,128,power_law_1.01,0.1045375943183899
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,2048,768,8,128,1,128,power_law_1.2,0.13049600124359131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,128,power_law_1.01,0.10657279491424561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,128,power_law_1.01,0.11161600351333618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,128,power_law_1.01,0.11783039569854736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,128,power_law_1.01,0.8853376388549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,128,power_law_1.01,0.12488960027694702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,128,power_law_1.01,0.13494399785995484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,128,power_law_1.01,0.14611200094223023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,128,power_law_1.01,1.2957183837890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,128,power_law_1.01,0.15866880416870116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,128,power_law_1.01,0.18384640216827391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,128,power_law_1.01,1.6724544525146485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,128,power_law_1.01,0.23092479705810548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,128,power_law_1.01,0.28159360885620116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2048,768,8,128,1,128,power_law_1.2,0.8092543601989746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,128,power_law_1.01,0.39054720401763915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,128,balanced,0.07338666419188182
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,128,balanced,0.07492800056934357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,128,balanced,0.08172266681989034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,128,power_law_1.01,2.5454080581665037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2048,768,8,128,1,128,power_law_1.2,0.8727871894836425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,128,power_law_1.01,3.3052928924560545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,2048,768,8,128,1,128,power_law_1.2,0.16408319473266603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,128,balanced,0.08737066388130188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,128,balanced,0.09159466624259949
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,128,balanced,0.10005333026250203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,128,balanced,0.1107360025246938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,128,balanced,0.12312533458073933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,128,balanced,0.1458186705907186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,128,balanced,0.1912053426106771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,128,balanced,0.2369813323020935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,128,balanced,0.33053332567214966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,128,balanced,0.4170986811319987
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,128,balanced,0.5902773141860962
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,128,balanced,0.7587146759033203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,128,power_law_1.01,0.49356799125671386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,128,balanced,1.0987520217895508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2048,768,8,128,1,128,power_law_1.2,1.2763456344604491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,128,power_law_1.01,0.7090879917144776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,128,balanced,1.4334719975789387
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,128,power_law_1.01,0.025510400533676147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,128,power_law_1.01,0.9259008407592774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,128,power_law_1.01,0.025651198625564576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,2048,768,8,128,1,128,power_law_1.2,1.1665472030639648
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,128,power_law_1.01,0.02558079957962036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,128,power_law_1.01,1.325228786468506
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,128,power_law_1.01,0.02771199941635132
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,128,power_law_1.01,0.02903040051460266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,128,power_law_1.01,0.03601279854774475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,128,power_law_1.01,1.7459392547607422
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,128,power_law_1.01,0.037241598963737486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,128,power_law_1.01,0.03898879885673523
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,128,power_law_1.01,0.04339199960231781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,128,power_law_1.01,0.04311679899692535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,128,power_law_1.01,0.039289599657058714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,128,power_law_1.01,2.5974912643432617
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,128,power_law_1.01,0.039961600303649904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,128,power_law_1.01,0.040166398882865904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,128,power_law_1.01,0.040320000052452086
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,128,power_law_1.01,0.042259201407432556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,128,power_law_1.01,0.042342400550842284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,128,power_law_1.01,3.5322113037109375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,128,power_law_1.01,0.04441600143909454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,1536,8,128,1,128,power_law_1.2,0.03359360098838806
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,128,power_law_1.01,0.04945279955863953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,1536,8,128,1,128,power_law_1.2,0.05913599729537964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,128,power_law_1.01,0.054502397775650024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,1536,8,128,1,128,power_law_1.2,0.05606399774551392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,128,power_law_1.01,0.0700160026550293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,1536,8,128,1,128,power_law_1.2,0.06504319906234741
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,128,power_law_1.01,0.08320639729499817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,1536,8,128,1,128,power_law_1.2,0.09203839898109437
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,128,power_law_1.01,0.11546880006790161
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,1536,8,128,1,128,power_law_1.2,0.09300479888916016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,128,power_law_1.01,0.14880640506744386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,128,power_law_1.01,0.21040639877319336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,128,balanced,2.1072266896565757
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,128,power_law_1.01,0.2720191955566406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,128,power_law_1.01,0.39594240188598634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,128,balanced,2.7869440714518228
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,128,power_law_1.01,0.5154751777648926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,1536,8,128,1,128,power_law_1.2,0.03587839901447296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,128,power_law_1.2,0.05359359979629517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,1536,8,128,1,128,power_law_1.2,0.09638400077819824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,1536,8,128,1,128,power_law_1.2,0.05633280277252197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,128,power_law_1.2,0.054176002740859985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,1536,8,128,1,128,power_law_1.2,0.09811199903488159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,128,power_law_1.2,0.056518399715423585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,1536,8,128,1,128,power_law_1.2,0.10131200551986694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,128,power_law_1.2,0.055904000997543335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,1536,8,128,1,128,power_law_1.2,0.1031999945640564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,128,power_law_1.2,0.06547840237617493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,1536,8,128,1,128,power_law_1.2,0.1080896019935608
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,128,power_law_1.2,0.06665599942207337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,1536,8,128,1,128,power_law_1.2,0.11386879682540893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,128,power_law_1.2,0.06740480065345764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,1536,8,128,1,128,power_law_1.2,0.12172800302505493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,128,power_law_1.2,0.0698303997516632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,1536,8,128,1,128,power_law_1.2,0.1342463970184326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,128,power_law_1.2,0.07362560033798218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,1536,8,128,1,128,power_law_1.2,0.14670720100402831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,128,power_law_1.2,0.07440000176429748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,1536,8,128,1,128,power_law_1.2,0.16171519756317138
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,1536,8,128,1,128,power_law_1.2,0.0572160005569458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,128,power_law_1.2,0.0800000011920929
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,1536,8,128,1,128,power_law_1.2,0.1842303991317749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,1536,8,128,1,128,power_law_1.2,0.06755840182304382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,1536,8,128,1,128,power_law_1.2,0.22763519287109374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,1536,8,128,1,128,power_law_1.2,0.09413120150566101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,1536,8,128,1,128,power_law_1.2,0.28140161037445066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,1536,8,128,1,128,power_law_1.2,0.09619839787483216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,1536,8,128,1,128,power_law_1.2,0.09909759759902954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,1536,8,128,1,128,power_law_1.2,0.40979838371276855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,1536,8,128,1,128,power_law_1.2,0.10103679895401001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,1536,8,128,1,128,power_law_1.2,0.5334527969360352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,1536,8,128,1,128,power_law_1.2,0.10423680543899536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,1536,8,128,1,128,power_law_1.2,0.10629760026931763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,1536,8,128,1,128,power_law_1.2,0.7372608184814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,1536,8,128,1,128,power_law_1.2,0.11150720119476318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,1536,8,128,1,128,power_law_1.2,0.11725440025329589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,128,power_law_1.2,0.08475520014762879
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,1536,8,128,1,128,power_law_1.2,0.1239743947982788
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,128,power_law_1.2,0.09513599872589111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,128,power_law_1.2,0.10584319829940796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,128,power_law_1.2,0.11754239797592163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,128,power_law_1.2,0.12929279804229737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,128,power_law_1.2,0.15544960498809815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,128,power_law_1.2,0.21149439811706544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,1536,8,128,1,128,power_law_1.01,0.049958398938179015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,128,power_law_1.2,0.26484479904174807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,1536,8,128,1,128,power_law_1.01,0.0542527973651886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,128,power_law_1.2,0.37029759883880614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,128,power_law_1.2,0.4807551860809326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,1536,8,128,1,128,power_law_1.2,0.1342463970184326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,1536,8,128,1,128,power_law_1.2,0.14431359767913818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,128,power_law_1.2,0.6707136154174804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,1536,8,128,1,128,power_law_1.2,0.15611519813537597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,1536,8,128,1,128,power_law_1.2,0.18188799619674684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,128,power_law_1.2,0.8767935752868652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,1536,8,128,1,128,power_law_1.2,0.22792959213256836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,1536,8,128,1,128,power_law_1.2,0.28643200397491453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,1536,8,128,1,128,power_law_1.01,0.05613440275192261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,128,power_law_1.2,1.3081343650817872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,1536,8,128,1,128,power_law_1.2,0.4057663917541504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,1536,8,128,1,128,power_law_1.01,0.05487359762191772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2048,768,8,128,1,128,power_law_1.2,1.2015487670898437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,1536,8,128,1,128,power_law_1.01,0.06643199920654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,1536,8,128,1,128,power_law_1.2,0.5229311943054199
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,1536,8,128,1,128,power_law_1.01,0.06549760103225707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,1536,8,128,1,128,power_law_1.01,0.06776959896087646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,1536,8,128,1,128,power_law_1.2,0.7723584175109863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,1536,8,128,1,128,power_law_1.01,0.070278400182724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,1536,8,128,1,128,power_law_1.2,0.9793279647827149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,1536,8,128,1,128,power_law_1.01,0.07408000230789184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,1536,8,128,1,128,power_law_1.01,0.0745024025440216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,1536,8,128,1,128,power_law_1.01,0.07976319789886474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,1536,8,128,1,128,power_law_1.2,1.4268480300903321
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,1536,8,128,1,128,power_law_1.01,0.08444799780845642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,1536,8,128,1,128,power_law_1.01,0.09185280203819275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,1536,8,128,1,128,power_law_1.01,0.10341119766235352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,1536,8,128,1,128,power_law_1.2,0.9916671752929688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,1536,8,128,1,128,power_law_1.01,0.1147007942199707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,1536,8,128,1,128,power_law_1.01,0.12691839933395385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,1536,8,128,1,128,power_law_1.01,0.15143040418624878
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,1536,8,128,1,128,power_law_1.01,0.19953279495239257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,1536,8,128,1,128,power_law_1.2,1.4680704116821288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,1536,8,128,1,128,power_law_1.01,0.25063679218292234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,1536,8,128,1,128,power_law_1.01,0.35761280059814454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,1536,8,128,1,128,power_law_1.01,0.4532288074493408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,1536,8,128,1,128,power_law_1.01,0.6450111865997314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,1536,8,128,1,128,power_law_1.01,0.824345588684082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,1536,8,128,1,128,power_law_1.01,1.2370816230773927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,1536,8,128,1,128,power_law_1.2,1.911756706237793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,128,power_law_1.01,1.5924991607666015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,2048,768,8,128,1,128,power_law_1.2,0.23850879669189454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,128,power_law_1.01,2.398508834838867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,128,power_law_1.01,3.128390312194824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,4096,1536,8,128,1,128,power_law_1.2,0.025887998938560485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,4096,1536,8,128,1,128,power_law_1.2,0.02550399899482727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,4096,1536,8,128,1,128,power_law_1.2,0.026047998666763307
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,4096,1536,8,128,1,128,power_law_1.2,0.027807998657226562
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,4096,1536,8,128,1,128,power_law_1.2,0.02908799946308136
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,4096,1536,8,128,1,128,power_law_1.2,0.03617919981479645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,4096,1536,8,128,1,128,power_law_1.2,0.037503999471664426
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,4096,1536,8,128,1,128,power_law_1.2,0.039315199851989745
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,4096,1536,8,128,1,128,power_law_1.2,0.043942400813102724
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,4096,1536,8,128,1,128,power_law_1.2,0.04314239919185638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,4096,1536,8,128,1,128,power_law_1.2,0.039417600631713866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,4096,1536,8,128,1,128,power_law_1.2,0.03989120125770569
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,4096,1536,8,128,1,128,power_law_1.2,0.04023039937019348
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,4096,1536,8,128,1,128,power_law_1.2,0.040678399801254275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,4096,1536,8,128,1,128,power_law_1.2,0.042444801330566405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,4096,1536,8,128,1,128,power_law_1.2,0.043007999658584595
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,4096,1536,8,128,1,128,power_law_1.2,0.045184001326560974
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,4096,1536,8,128,1,128,power_law_1.2,0.05046399831771851
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,4096,1536,8,128,1,128,power_law_1.2,0.0568448007106781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,4096,1536,8,128,1,128,power_law_1.2,0.07159039974212647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,1536,8,128,1,128,power_law_1.2,1.7260543823242187
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,4096,1536,8,128,1,128,power_law_1.2,0.09144960045814514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2048,768,8,128,1,128,power_law_1.2,1.693440055847168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,128,balanced,0.07844266792138417
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,128,balanced,0.07721066474914551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,128,balanced,0.13358400265375772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,128,balanced,0.14351466298103333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,128,balanced,0.17972799142201742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,128,balanced,0.2683466672897339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,128,balanced,0.3722933530807495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,128,balanced,0.37468798955281574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,1536,8,128,1,128,power_law_1.2,2.7631296157836913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,128,balanced,0.3797760009765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,128,balanced,0.38497598965962726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,128,balanced,0.39215465386708576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,128,balanced,0.4004266659418742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,128,balanced,0.4084320068359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,2048,768,8,128,1,128,power_law_1.2,0.31434879302978513
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,128,balanced,0.4224373499552409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,128,balanced,0.05014933149019877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,128,balanced,0.04608533283074697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,128,balanced,0.4392000039418538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,128,balanced,0.08281066517035167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,128,balanced,0.10021332899729411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,128,balanced,0.45584531625111896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,128,balanced,0.138154665629069
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,128,balanced,0.25682133436203003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,128,balanced,0.48498133818308514
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,4096,1536,8,128,1,128,power_law_1.2,0.1291327953338623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,128,balanced,0.375055988629659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,128,balanced,0.545632004737854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,128,balanced,0.37885868549346924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,128,balanced,0.38151466846466064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,128,balanced,0.38674668471018475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,128,balanced,0.39634132385253906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,128,balanced,0.4031360149383545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,128,balanced,0.41230932871500653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,4096,1536,8,128,1,128,power_law_1.2,0.16780799627304077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,128,balanced,0.4248480002085368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,128,balanced,0.44274131457010907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,128,balanced,0.45948266983032227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,128,balanced,0.49224531650543213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,128,balanced,0.5573279857635498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,128,balanced,0.6208479801813761
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,128,balanced,0.6069013277689616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,128,balanced,0.7487413088480631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,128,balanced,0.7290240128835043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,128,balanced,0.8747519652048746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,128,balanced,0.851967970530192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,128,balanced,1.1256639957427979
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,128,balanced,1.0957067012786865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,128,balanced,1.3919199307759602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,128,balanced,1.3474292755126953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,128,balanced,1.8312853177388508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,128,balanced,2.340325355529785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,128,balanced,3.3584906260172525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,128,balanced,1.8900426228841145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,1536,8,128,1,128,power_law_1.2,2.587001609802246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,128,balanced,2.41159995396932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,128,balanced,4.463712056477864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,128,balanced,0.07060266534487407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,128,balanced,0.07483200232187907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,128,balanced,0.08030933141708374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,128,balanced,0.0811466674009959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,128,balanced,0.09574400385220845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,128,balanced,0.1523413360118866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,128,balanced,3.4396371841430664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,128,balanced,0.2197706699371338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,128,balanced,0.22132267554601034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,128,balanced,0.22604266802469888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,128,balanced,0.22819199164708456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,1536,8,128,1,128,power_law_1.2,3.9256256103515623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,128,balanced,0.23657600084940592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,128,balanced,0.24612800280253092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,128,balanced,0.25243733326594037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,128,balanced,0.26746666431427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,128,balanced,4.513552029927571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,128,balanced,0.2834879954655965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,4096,1536,8,128,1,128,power_law_1.2,0.2332159996032715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,128,balanced,0.29850133260091144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,128,balanced,0.3312159975369771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,128,balanced,0.39181868235270184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,128,balanced,0.4517333507537842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,128,balanced,0.572378675142924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,1536,8,128,1,128,power_law_1.2,1.9646207809448242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,128,balanced,0.6905439694722494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,128,balanced,0.9299200375874838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,128,balanced,1.18177064259847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,128,balanced,1.652880032857259
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,128,balanced,2.147530714670817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,128,power_law_1.01,0.09701120257377624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,128,balanced,3.1083412170410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,4096,1536,8,128,1,128,power_law_1.2,0.3040640115737915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,128,balanced,0.029333333174387615
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,128,balanced,0.029343999922275543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,128,balanced,0.030063999195893604
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,128,balanced,0.03136533250411352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,128,balanced,0.033802665770053864
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,128,balanced,0.037802666425704956
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,128,balanced,0.05514666438102722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,128,balanced,4.089994748433431
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,128,balanced,0.05942399799823761
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,128,balanced,0.10905599594116211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,128,balanced,0.10956266522407532
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,128,balanced,0.1077280044555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,128,balanced,0.09288533528645833
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,128,balanced,0.08083733419577281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,128,balanced,0.0647680014371872
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,128,balanced,0.059445331494013466
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,128,balanced,0.0617386649052302
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,128,balanced,0.0665280024210612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,128,balanced,0.07464000085989635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,128,balanced,0.08555733164151509
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,128,balanced,0.10534399747848511
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,128,balanced,0.12590932846069336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,128,balanced,0.1657813290754954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,128,balanced,0.207696000734965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,128,balanced,0.2916853427886963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,128,power_law_1.01,0.09673600196838379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,128,power_law_1.01,0.155731201171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,128,power_law_1.01,0.1609984040260315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,128,power_law_1.01,0.19614720344543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,128,power_law_1.01,0.08610560297966004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,128,power_law_1.01,0.2678976058959961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,128,power_law_1.01,0.08655359745025634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,128,balanced,0.38364799817403156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,128,power_law_1.01,0.37148160934448243
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,128,balanced,0.539189338684082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,128,power_law_1.01,0.12263679504394531
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,128,balanced,0.7071200211842855
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,128,power_law_1.01,0.3743488073348999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,128,power_law_1.01,0.08353279829025269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,128,power_law_1.01,0.3784064054489136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,1536,8,128,1,128,power_law_1.2,2.8717119216918947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,128,power_law_1.01,0.08663679957389832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,128,power_law_1.01,0.38419840335845945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,128,power_law_1.01,0.09598079919815064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,128,power_law_1.01,0.39225599765777586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,128,power_law_1.01,0.0970687985420227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,128,power_law_1.01,0.39945600032806394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,128,power_law_1.01,0.11291520595550537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,128,power_law_1.01,0.1332800030708313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,128,power_law_1.01,0.40828800201416016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,128,power_law_1.01,0.15598080158233643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,128,power_law_1.01,0.16684800386428833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,128,power_law_1.01,0.4225344181060791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,128,power_law_1.01,0.21592319011688232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,128,power_law_1.01,0.2579263925552368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,128,power_law_1.01,0.43929600715637207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,128,power_law_1.01,0.21984000205993653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,128,power_law_1.01,0.3709183931350708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,128,power_law_1.01,0.45372800827026366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,128,power_law_1.01,0.2227328062057495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,128,power_law_1.01,0.37426559925079345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,128,power_law_1.01,0.48774399757385256
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,128,power_law_1.01,0.22855041027069092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,128,power_law_1.01,0.37946879863739014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,128,power_law_1.01,0.5512256145477294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,128,power_law_1.01,0.23664638996124268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,128,power_law_1.01,0.3834304094314575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,128,power_law_1.01,0.6157951831817627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,128,power_law_1.01,0.2437119960784912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,128,power_law_1.01,0.2514559984207153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,128,power_law_1.01,0.7619328022003173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,128,power_law_1.01,0.26578559875488283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,128,power_law_1.01,0.9051456451416016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,128,power_law_1.01,0.282092809677124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,128,power_law_1.01,0.29482879638671877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,128,power_law_1.01,1.2145983695983886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,128,power_law_1.01,0.3279040098190308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,128,power_law_1.01,0.3916991949081421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,128,power_law_1.01,1.5206975936889648
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,128,power_law_1.01,0.3984575986862183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,128,power_law_1.01,0.4080959796905518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,128,power_law_1.01,2.0557376861572267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,128,power_law_1.01,0.4220416069030762
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,128,power_law_1.01,0.4393472194671631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,128,power_law_1.01,0.39967999458312986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,128,power_law_1.01,2.686771202087402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,128,power_law_1.01,0.4552000045776367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,128,power_law_1.01,0.4689727783203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,128,power_law_1.01,0.49008641242980955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,128,power_law_1.01,0.6039040088653564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,128,power_law_1.01,0.5585855960845947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,128,power_law_1.01,0.7350080013275146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,128,power_law_1.01,0.625600004196167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,128,power_law_1.01,3.915340805053711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,128,power_law_1.01,1.0027584075927733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,128,power_law_1.01,1.2698623657226562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,128,power_law_1.01,5.209651184082031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,128,power_law_1.01,1.8139520645141602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,128,power_law_1.01,0.7636352062225342
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,128,power_law_1.01,0.9216064453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,128,power_law_1.01,2.362803268432617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,128,power_law_1.01,1.2188799858093262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,128,power_law_1.01,1.521401596069336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,128,power_law_1.01,3.3879425048828127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,128,power_law_1.01,2.1146303176879884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,128,power_law_1.01,4.449151992797852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,128,power_law_1.01,2.6735488891601564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,384,1,128,power_law_1.2,0.09711359739303589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,384,1,128,power_law_1.2,0.09651200175285339
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,384,1,128,power_law_1.2,0.15555200576782227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,384,1,128,power_law_1.2,0.1622015953063965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,128,power_law_1.01,0.029702401161193846
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,128,power_law_1.01,3.9469886779785157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,128,power_law_1.01,0.02998400032520294
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,128,power_law_1.01,0.03079040050506592
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,128,power_law_1.01,0.03208959996700287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,128,power_law_1.01,0.03384959995746613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,2048,768,8,128,1,128,power_law_1.2,1.525657558441162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,128,power_law_1.01,5.154899215698242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,384,1,128,power_law_1.2,0.19541120529174805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,384,1,128,power_law_1.2,0.08528000116348267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,384,1,128,power_law_1.2,0.08465920090675354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,384,1,128,power_law_1.2,0.08541439771652222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,384,1,128,power_law_1.2,0.26686079502105714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,384,1,128,power_law_1.2,0.08820480108261108
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,384,1,128,power_law_1.2,0.12249599695205689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,384,1,128,power_law_1.2,0.37087359428405764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,384,1,128,power_law_1.2,0.09620479941368103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,384,1,128,power_law_1.2,0.13267840147018434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,384,1,128,power_law_1.2,0.37416958808898926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,384,1,128,power_law_1.2,0.0972544014453888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,384,1,128,power_law_1.2,0.1638335943222046
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,128,power_law_1.01,0.03952000141143799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,384,1,128,power_law_1.2,0.11382399797439575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,384,1,128,power_law_1.2,0.2556544065475464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,128,power_law_1.01,0.049830400943756105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,384,1,128,power_law_1.2,0.1541375994682312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,384,1,128,power_law_1.2,0.36937599182128905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,128,power_law_1.01,0.05932160019874573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,384,1,128,power_law_1.2,0.21249279975891114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,384,1,128,power_law_1.2,0.3746623992919922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,128,power_law_1.01,0.10391680002212525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,384,1,128,power_law_1.2,0.2182528018951416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,384,1,128,power_law_1.2,0.37845759391784667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,384,1,128,power_law_1.2,0.22320640087127686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,384,1,128,power_law_1.2,0.37760000228881835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,384,1,128,power_law_1.2,0.3832832098007202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,384,1,128,power_law_1.2,0.22817280292510986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,384,1,128,power_law_1.2,0.39071359634399416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,384,1,128,power_law_1.2,0.23626880645751952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,384,1,128,power_law_1.2,0.39850239753723143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,384,1,128,power_law_1.2,0.2422719955444336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,384,1,128,power_law_1.2,0.40672640800476073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,384,1,128,power_law_1.2,0.2514240026473999
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,128,power_law_1.01,0.10273280143737792
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,384,1,128,power_law_1.2,0.4207615852355957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,384,1,128,power_law_1.2,0.2634687900543213
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,128,power_law_1.01,0.10156160593032837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,384,1,128,power_law_1.2,0.4373311996459961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,384,1,128,power_law_1.2,0.28149120807647704
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,128,power_law_1.01,0.08187519907951354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,384,1,128,power_law_1.2,0.4537024021148682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,384,1,128,power_law_1.2,0.30053119659423827
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,128,power_law_1.01,0.07274240255355835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,384,1,128,power_law_1.2,0.49181442260742186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,4096,1536,8,128,1,128,power_law_1.2,0.4461631774902344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,128,power_law_1.01,0.059564799070358276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,384,1,128,power_law_1.2,0.5613696098327636
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,128,power_law_1.01,0.05653759837150574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,384,1,128,power_law_1.2,0.6282432079315186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,128,power_law_1.01,0.05653759837150574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,128,power_law_1.01,0.06410239934921265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,384,1,128,power_law_1.2,0.7868031978607177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,384,1,128,power_law_1.2,0.3834559917449951
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,128,power_law_1.01,0.07244799733161926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,384,1,128,power_law_1.2,0.3883392095565796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,384,1,128,power_law_1.2,0.3355007886886597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,384,1,128,power_law_1.2,0.39944961071014407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,384,1,128,power_law_1.2,0.40716161727905276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,384,1,128,power_law_1.2,0.4221183776855469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,384,1,128,power_law_1.2,0.43975038528442384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,384,1,128,power_law_1.2,0.9446656227111816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,384,1,128,power_law_1.2,0.4547391891479492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,384,1,128,power_law_1.2,0.40529279708862304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,384,1,128,power_law_1.2,0.48956799507141113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,384,1,128,power_law_1.2,1.2746496200561523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,384,1,128,power_law_1.2,0.4753727912902832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,384,1,128,power_law_1.2,0.554643201828003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,384,1,128,power_law_1.2,0.6161856174468994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,384,1,128,power_law_1.2,1.5826432228088378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,384,1,128,power_law_1.2,0.7535808086395264
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,128,power_law_1.01,0.08505600094795226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,384,1,128,power_law_1.2,1.039583969116211
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,128,power_law_1.01,0.10936319828033447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,384,1,128,power_law_1.2,2.2836671829223634
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,128,power_law_1.01,0.13352320194244385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,384,1,128,power_law_1.2,1.3365823745727539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,128,power_law_1.01,0.17779200077056884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,384,1,128,power_law_1.2,0.6192383766174316
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,128,power_law_1.01,0.22539520263671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,384,1,128,power_law_1.2,3.10263671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,384,1,128,power_law_1.2,0.791916799545288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,384,1,128,power_law_1.2,1.872332763671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,384,1,128,power_law_1.2,0.9364992141723633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,384,1,128,power_law_1.2,2.5091968536376954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,384,1,128,power_law_1.2,4.182400131225586
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,128,power_law_1.01,0.33075199127197263
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,128,power_law_1.01,0.42137598991394043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,384,1,128,power_law_1.2,3.5237823486328126
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,128,power_law_1.01,0.6151552200317383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,384,1,128,power_law_1.2,1.2745856285095214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,384,1,128,power_law_1.2,5.5499519348144535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,128,power_law_1.01,0.811302375793457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,128,balanced,0.03382933388153712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,384,1,128,power_law_1.2,1.5625215530395509
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,128,balanced,0.03246400008598963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,128,balanced,0.052671998739242554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,128,balanced,0.055733333031336464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,128,balanced,0.08063999811808269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,128,balanced,0.1441333293914795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,128,balanced,0.14660799503326416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,128,balanced,0.14860799908638
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,128,balanced,0.1523253321647644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,128,balanced,0.1532693306605021
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,128,balanced,0.15846400459607443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,384,1,128,power_law_1.2,2.2417343139648436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,128,balanced,0.1646346648534139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,128,balanced,0.1686826745669047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,128,balanced,0.1781066656112671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,128,balanced,0.1877280076344808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,128,balanced,0.19732799132665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,128,balanced,0.21804799636205038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,128,balanced,0.2561066746711731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,128,balanced,0.2956213355064392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,128,balanced,0.37664000193277997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,384,1,128,power_law_1.2,2.941152000427246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,128,balanced,0.4562026659647624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,128,balanced,0.6118293205897013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,128,balanced,0.7606346607208252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,128,balanced,1.0628693103790283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,128,balanced,1.3629333178202312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,384,1,128,power_law_1.2,4.098681640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,128,balanced,1.9605387051900227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,384,1,128,power_law_1.2,4.761779022216797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,128,balanced,0.0376800000667572
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,128,balanced,0.037621334195137024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,128,balanced,0.052015999952952065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,128,balanced,2.5735839207967124
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,128,balanced,0.05602666735649109
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,384,1,128,power_law_1.2,0.03041279911994934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,128,balanced,0.08578133583068848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,128,balanced,0.14618133505185446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,128,balanced,0.1504639983177185
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,384,1,128,power_law_1.2,0.02974080145359039
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,384,1,128,power_law_1.2,0.03057279884815216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,384,1,128,power_law_1.2,5.623596954345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,384,1,128,power_law_1.2,0.032102400064468385
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,384,1,128,power_law_1.2,0.03434880077838898
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,384,1,128,power_law_1.2,0.03852159976959228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,128,balanced,0.048309331138928734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,128,balanced,0.0509493350982666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,128,balanced,0.05123733480771383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,128,balanced,0.05153066913286845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,128,balanced,0.1506666640440623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,128,balanced,0.05619733532269796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,128,balanced,0.15454399585723877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,128,balanced,0.09085866808891296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,128,balanced,0.15665066242218018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,128,balanced,0.09307199716567993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,128,balanced,0.16261866688728333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,128,balanced,0.16675732533137003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,128,balanced,0.170522669951121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,128,balanced,0.18098666270573935
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,128,balanced,0.191210667292277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,128,balanced,0.19940799474716187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,128,balanced,0.2206719915072123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,128,balanced,0.26226667563120526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,1536,8,128,1,128,power_law_1.2,3.4209663391113283
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,128,balanced,0.02515200028816859
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,128,balanced,0.025386666258176167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,128,power_law_1.01,0.03387520015239716
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,128,balanced,0.025205334027608235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,128,balanced,0.025093334416548412
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,128,balanced,0.0271519993742307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,128,power_law_1.01,0.03514240086078644
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,384,1,128,power_law_1.2,0.05125759840011597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,128,power_law_1.01,0.058508801460266116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,128,balanced,0.0930560032526652
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,384,1,128,power_law_1.2,0.055379199981689456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,128,balanced,0.09709333380063374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,128,balanced,0.09723732868830363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,128,power_law_1.01,0.06128000020980835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,128,balanced,0.10526399811108907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,128,balanced,0.10801600416501363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,128,balanced,0.11154133081436157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,384,1,128,power_law_1.2,0.09357439875602722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,128,balanced,0.12372799714406331
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,128,balanced,0.13205333550771078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,128,balanced,0.14220800002415976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,128,balanced,0.16425066192944845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,128,balanced,0.20442666610081991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,128,balanced,0.24471465746561685
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,128,balanced,0.32785600423812866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,128,balanced,0.4086560010910034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,128,balanced,0.5713866551717123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,128,balanced,0.7234186331431071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,128,balanced,1.0356746514638264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,128,power_law_1.01,0.08177919983863831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,128,power_law_1.01,0.14304640293121337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,128,balanced,1.3451520601908367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,128,balanced,0.30427734057108563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,128,balanced,0.3885173400243123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,128,balanced,0.4712800184885661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,128,balanced,1.958624045054118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,128,balanced,0.6318666537602743
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,128,balanced,0.02994133283694585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,128,balanced,0.03939199944337209
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,128,balanced,0.03977066775163015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,128,balanced,0.7875253359476725
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,128,balanced,0.03986666599909464
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,128,balanced,0.03974399964014689
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,128,balanced,0.05180266499519348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,128,balanced,1.0963679949442546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,128,balanced,1.4058133761088054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,384,1,128,power_law_1.2,0.09832320213317872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,128,power_law_1.01,0.14617600440979003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,128,balanced,2.03275203704834
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,384,1,128,power_law_1.2,0.09687039852142335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,128,power_law_1.01,0.14750080108642577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,128,power_law_1.01,0.15072640180587768
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,384,1,128,power_law_1.2,0.08417279720306396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,128,power_law_1.01,0.15140479803085327
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,384,1,128,power_law_1.2,0.06667519807815551
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,128,balanced,0.03974399964014689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,128,power_law_1.01,0.15653760433197023
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,384,1,128,power_law_1.2,0.06270080208778381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,128,power_law_1.01,0.16206079721450806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,1536,8,128,1,128,power_law_1.2,3.843385696411133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,128,power_law_1.01,0.1664191961288452
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,384,1,128,power_law_1.2,0.0545087993144989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,128,power_law_1.01,0.17601280212402343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,128,power_law_1.01,0.033792001008987424
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,384,1,128,power_law_1.2,0.05862399935722351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,128,power_law_1.01,0.18739839792251586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,128,power_law_1.01,0.03640959858894348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,128,power_law_1.01,0.19623039960861205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,128,power_law_1.01,0.04970879852771759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,128,power_law_1.01,0.2169856071472168
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,128,power_law_1.01,0.05609599947929382
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,128,balanced,2.5832799275716147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,128,power_law_1.01,0.25655040740966795
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,128,balanced,0.04021333406368891
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,128,balanced,0.039503999054431915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,128,power_law_1.01,0.0846783995628357
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,128,balanced,0.04090133309364319
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,128,power_law_1.01,0.29752960205078127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,128,balanced,0.0418453315893809
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,128,balanced,0.04385599990685781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,128,power_law_1.01,0.14501760005950928
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,128,balanced,0.04563733438650767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,128,power_law_1.01,0.38154239654541017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,128,balanced,0.04796266555786133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,128,power_law_1.01,0.14803199768066405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,128,power_law_1.01,0.4664639949798584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,128,power_law_1.01,0.1493888020515442
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,384,1,128,power_law_1.2,0.06353920102119445
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,128,power_law_1.01,0.15268479585647582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,384,1,128,power_law_1.2,0.07694720029830933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,128,power_law_1.01,0.1542400002479553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,128,power_law_1.01,0.15870720148086548
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,128,power_law_1.01,0.048825600743293764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,128,power_law_1.01,0.16405760049819945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,128,power_law_1.01,0.04954879879951477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,128,power_law_1.01,0.16862720251083374
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,4096,1536,8,128,1,128,power_law_1.2,0.5882688045501709
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,128,balanced,0.056176001826922096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,128,power_law_1.01,0.17895040512084961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,128,power_law_1.01,0.05356799960136414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,128,power_law_1.01,0.18858879804611206
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,128,power_law_1.01,0.053830397129058835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,128,power_law_1.01,0.6284416198730469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,128,power_law_1.01,0.05909119844436646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,128,balanced,2.654575983683268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,128,power_law_1.01,0.08842880129814149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,128,power_law_1.2,0.03449600040912628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,384,1,128,power_law_1.2,0.09425280094146729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,128,power_law_1.01,0.09128959774971009
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,128,power_law_1.2,0.03468799889087677
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,384,1,128,power_law_1.2,0.11475839614868164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,128,power_law_1.01,0.09286400079727172
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,384,1,128,power_law_1.2,0.13518719673156737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,128,power_law_1.01,0.09601920247077941
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,128,balanced,0.06434666613737743
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,128,balanced,0.0825493335723877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,384,1,128,power_law_1.2,0.1864575982093811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,128,power_law_1.01,0.09702399969100953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,128,power_law_1.01,0.19812480211257935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,384,1,128,power_law_1.2,0.24240000247955323
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,128,power_law_1.01,0.21852800846099854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,128,power_law_1.01,0.025414401292800905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,128,power_law_1.01,0.8020480155944825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,128,power_law_1.01,0.26238079071044923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,128,power_law_1.01,0.025094398856163026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,128,power_law_1.01,0.30215680599212646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,128,power_law_1.01,0.02508159875869751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,128,power_law_1.2,0.05633919835090637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,128,power_law_1.01,0.383622407913208
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,128,power_law_1.01,0.025363200902938844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,128,power_law_1.2,0.06348159909248352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,128,power_law_1.01,0.47457280158996584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,128,balanced,0.10260267059008281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,128,balanced,0.1520799994468689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,128,power_law_1.01,0.10227839946746826
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,128,balanced,0.19549866517384848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,128,power_law_1.01,0.6531392097473144
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,128,balanced,0.27937599023183185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,128,power_law_1.01,0.10720000267028809
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,384,1,128,power_law_1.2,0.3445120096206665
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,128,power_law_1.01,0.8309696197509766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,384,1,128,power_law_1.2,0.4509119987487793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,128,power_law_1.01,1.140659236907959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,128,power_law_1.01,1.1632255554199218
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,384,1,128,power_law_1.2,0.6515711784362793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,128,power_law_1.01,0.027507200837135315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,128,power_law_1.01,1.4908096313476562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,128,power_law_1.01,1.5175104141235352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,128,power_law_1.01,0.029631999135017396
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,384,1,128,power_law_1.2,0.9028096199035645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,128,power_law_1.01,0.03758080005645752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,128,balanced,0.36109864711761475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,128,power_law_1.2,0.034534400701522826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,128,power_law_1.01,0.11147520542144776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,128,power_law_1.01,2.1689216613769533
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,128,power_law_1.01,0.0389631986618042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,128,power_law_1.2,0.04878720045089722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,128,power_law_1.2,0.03601920008659363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,128,power_law_1.01,2.1799936294555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,128,power_law_1.01,0.03934719860553741
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,128,power_law_1.2,0.049702399969100954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,128,power_law_1.01,0.04000000059604645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,128,power_law_1.2,0.05370240211486817
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,128,power_law_1.01,0.04798719882965088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,128,power_law_1.2,0.05246080160140991
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,128,power_law_1.2,0.08265600204467774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,128,power_law_1.01,0.03902080059051514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,128,power_law_1.2,0.058380800485610965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,128,power_law_1.2,0.14273279905319214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,128,power_law_1.2,0.08870400190353393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,128,power_law_1.2,0.14565759897232056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,128,power_law_1.2,0.09127680063247681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,128,power_law_1.2,0.14737919569015503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,128,power_law_1.01,0.12151679992675782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,128,power_law_1.2,0.09287040233612061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,128,power_law_1.2,0.1507840037345886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,128,power_law_1.01,0.13256959915161132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,128,power_law_1.2,0.09569919705390931
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,128,power_law_1.01,0.14387840032577515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,128,power_law_1.2,0.15221760272979737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,128,power_law_1.2,0.0972544014453888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,128,power_law_1.2,0.1565440058708191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,128,power_law_1.01,2.8211200714111326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,128,power_law_1.2,0.10234880447387695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,128,power_law_1.2,0.16176639795303344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,128,power_law_1.01,0.03950720131397247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,128,power_law_1.2,0.10720640420913696
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,128,power_law_1.2,0.16598399877548217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,128,power_law_1.01,0.03912320137023926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,128,power_law_1.2,0.1118016004562378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,128,power_law_1.2,0.1756608009338379
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,128,power_law_1.01,0.03966720104217529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,128,power_law_1.2,0.12209919691085816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,128,power_law_1.2,0.18743679523468018
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,128,power_law_1.01,0.041094401478767396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,128,power_law_1.2,0.1336192011833191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,128,power_law_1.2,0.19703680276870728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,128,power_law_1.2,0.049830400943756105
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,128,power_law_1.01,0.04229759871959686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,128,power_law_1.2,0.14454400539398193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,128,power_law_1.2,0.21850240230560303
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,128,power_law_1.2,0.05607039928436279
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,128,power_law_1.01,0.045542401075363156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,128,power_law_1.2,0.2579583883285522
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,128,power_law_1.01,0.16482559442520142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,128,power_law_1.01,0.048582398891448976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,128,power_law_1.2,0.30017919540405275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,128,power_law_1.01,0.2083967924118042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,128,power_law_1.01,2.919552040100098
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,128,balanced,0.03329599897066752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,128,power_law_1.2,0.3905855894088745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,128,power_law_1.2,0.4984640121459961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,128,power_law_1.2,0.02508159875869751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,128,power_law_1.2,0.6792704105377197
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,128,power_law_1.2,0.02489600032567978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,128,power_law_1.2,0.16662399768829345
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,128,power_law_1.2,0.02470400035381317
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,128,power_law_1.2,0.08487679958343505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,128,power_law_1.2,0.21172480583190917
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,128,power_law_1.2,0.025407999753952026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,128,power_law_1.2,0.1451583981513977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,128,power_law_1.2,0.25579519271850587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,128,power_law_1.2,0.027059200406074523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,128,power_law_1.01,0.2514688014984131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,128,power_law_1.2,0.14791040420532225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,128,power_law_1.2,0.3447103977203369
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,128,power_law_1.2,0.029951998591423036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,128,power_law_1.2,0.14984960556030275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,128,power_law_1.2,0.4362175941467285
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,128,power_law_1.2,0.03740800023078918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,128,power_law_1.2,0.1527999997138977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2048,768,8,128,1,128,power_law_1.2,1.5938112258911132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,128,power_law_1.2,0.6097216129302978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,128,balanced,0.033861334125200905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,128,balanced,0.033813332517941795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,128,balanced,0.051541333397229515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,128,balanced,0.05820799867312113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,128,power_law_1.2,0.8603903770446777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,128,balanced,0.08469866712888081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,128,balanced,0.14940800269444784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,128,balanced,0.15011733770370483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,128,power_law_1.01,0.060198402404785155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,128,balanced,0.15030399958292642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,128,balanced,0.15425599614779154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,128,balanced,0.15433067083358765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,128,balanced,0.031898667414983116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,128,balanced,0.1605280041694641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,128,balanced,0.05225066840648651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,128,balanced,0.1665600041548411
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,128,balanced,0.16895999511082968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,128,power_law_1.01,0.34250879287719727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,128,balanced,0.18105065822601318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,128,balanced,0.18951465686162314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,128,balanced,0.20149866739908853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,128,power_law_1.01,0.42682881355285646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,128,balanced,0.21988266706466675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,128,balanced,0.26255999008814496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,128,power_law_1.2,0.03906559944152832
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,128,power_law_1.2,0.15445760488510132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,128,balanced,0.3038506706555684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,128,power_law_1.01,0.5987199783325196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,128,balanced,0.3866186539332072
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,128,power_law_1.2,0.039577600359916684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,128,power_law_1.2,0.78373122215271
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,128,power_law_1.01,0.7607679843902588
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,128,power_law_1.01,0.07046399712562561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,128,power_law_1.2,1.119001579284668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,128,balanced,0.054645334680875145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,128,power_law_1.01,0.09274880290031433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,128,balanced,0.08269333342711131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,128,power_law_1.01,1.090003204345703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,128,balanced,0.14405333002408346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,128,balanced,0.146314670642217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,128,power_law_1.01,0.11918079853057861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,128,balanced,0.14830933014551798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,128,balanced,0.15240533153216043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,128,balanced,0.15436266859372458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,128,power_law_1.2,1.4801024436950683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,128,balanced,0.15653333067893982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,128,power_law_1.01,1.4319040298461914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,128,power_law_1.2,0.15918079614639283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,128,balanced,0.4699893395105998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,128,power_law_1.2,0.1649664044380188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,128,power_law_1.2,0.04012160003185272
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,128,power_law_1.2,2.1845312118530273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,128,power_law_1.2,0.16848000288009643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,128,power_law_1.2,1.1885184288024901
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,128,power_law_1.2,0.04746879935264588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,128,power_law_1.2,0.17868800163269044
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,128,power_law_1.2,0.03943679928779602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,128,power_law_1.2,0.1890944004058838
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,128,power_law_1.2,0.039827200770378116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,128,power_law_1.2,0.19778560400009154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,128,power_law_1.01,0.170579195022583
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,128,power_law_1.2,0.039136001467704774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,128,power_law_1.2,0.21986560821533202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,128,balanced,0.16288533806800842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,128,balanced,0.16661333044370016
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,128,power_law_1.01,0.21918718814849852
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,128,balanced,0.17683732509613037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,128,power_law_1.2,0.2606015920639038
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,128,balanced,0.18805867433547974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,128,balanced,0.19726399580637613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,128,power_law_1.01,0.3148799896240234
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,128,balanced,0.21651200453440347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,128,balanced,0.6286186774571737
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,128,balanced,0.25434666872024536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,128,balanced,0.2946026722590129
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,128,power_law_1.01,0.4091328144073486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,128,balanced,0.7815252939860026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,128,balanced,0.374506672223409
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,128,balanced,0.45626668135325116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,128,balanced,1.094821294148763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,128,balanced,0.6092160145441691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,128,power_law_1.2,1.5640576362609864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,128,balanced,0.7593546708424886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,128,power_law_1.2,2.85677433013916
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,128,power_law_1.2,0.03986560106277466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,128,balanced,1.0565919876098633
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,128,power_law_1.2,0.04013440012931824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,128,power_law_1.2,0.3019392013549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,128,power_law_1.2,0.04325119853019714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,128,power_law_1.01,2.088435173034668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,128,power_law_1.2,0.3915519952774048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,128,balanced,0.04710400104522705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,128,balanced,0.050144001841545105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,128,balanced,0.052416001756985985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,128,power_law_1.2,0.49264001846313477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,128,balanced,0.052469333012898765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,128,balanced,0.056847999493281044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,128,balanced,0.09076799949010213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,128,balanced,0.09309333562850952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,128,balanced,0.09356266260147095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,128,power_law_1.2,0.6821760177612305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,128,balanced,0.09681066870689392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,128,balanced,0.09694400429725647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,128,power_law_1.01,2.7146175384521483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,128,balanced,0.10434133807818095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,128,balanced,0.10910933216412862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,128,balanced,0.11142399907112122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,128,power_law_1.2,0.8681856155395508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,128,balanced,0.123471995194753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,128,balanced,0.1320266624291738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,128,balanced,1.3579626083374023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,128,balanced,0.14336533347765604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,128,balanced,0.16250666975975037
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,128,balanced,0.025199999411900837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,128,balanced,0.20431999365488687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,128,power_law_1.2,0.04618239998817444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,128,power_law_1.2,1.2207743644714355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,128,balanced,0.2435306708017985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,128,balanced,1.9533813794453938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,128,balanced,0.3269279996554057
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,128,power_law_1.2,0.049830400943756105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,128,balanced,0.4095040162404378
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,128,power_law_1.2,0.06264960169792175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,128,balanced,0.5699199835459391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,128,power_law_1.2,1.5957375526428224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,128,balanced,1.3988000551859539
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,128,power_law_1.2,0.07377279996871948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,128,balanced,0.7239147027333578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,128,power_law_1.2,0.10138239860534667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,128,balanced,1.029317299524943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,128,power_law_1.01,0.034771201014518735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,128,balanced,2.018218676249186
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,128,power_law_1.2,0.12684160470962524
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,128,power_law_1.2,2.2324928283691405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,128,power_law_1.01,0.03445119857788086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,128,power_law_1.2,2.395699119567871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,128,balanced,1.3422773679097493
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,128,power_law_1.2,0.1852031946182251
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,128,balanced,0.02518933266401291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,128,power_law_1.01,0.05793280005455017
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,128,power_law_1.2,0.24198400974273682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,128,balanced,2.644261360168457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,128,power_law_1.01,0.061343997716903687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,128,power_law_1.2,0.3488192081451416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,128,power_law_1.01,0.08232319951057435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,128,power_law_1.2,3.063724708557129
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,128,balanced,2.5677919387817383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,128,power_law_1.2,3.1134719848632812
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,128,power_law_1.2,0.4644480228424072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,128,power_law_1.01,0.14350719451904298
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,128,power_law_1.01,0.048876801133155824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,128,power_law_1.2,0.03368319869041443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,128,power_law_1.2,0.034745600819587705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,128,power_law_1.01,0.14574079513549804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,128,power_law_1.01,0.05075200200080872
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,128,power_law_1.2,0.03407360017299652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,128,power_law_1.2,0.03591679930686951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,128,power_law_1.01,0.14732160568237304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,128,power_law_1.01,0.05384960174560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,128,power_law_1.2,0.057599997520446776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,128,power_law_1.2,0.051072001457214355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,128,power_law_1.01,0.15070079565048217
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,128,balanced,0.026693334182103474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,128,power_law_1.01,0.05416319966316223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,128,power_law_1.2,0.0559935986995697
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,128,power_law_1.01,0.15263999700546266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,128,balanced,1.957680066426595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,128,power_law_1.01,0.05907840132713318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,128,power_law_1.2,0.08458240032196045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,128,power_law_1.01,0.03464959859848023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,128,power_law_1.01,0.15676159858703614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,128,power_law_1.01,0.0889087975025177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,128,power_law_1.01,0.16174720525741576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,128,power_law_1.01,0.03624320030212402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,128,power_law_1.01,0.09117439985275269
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,128,power_law_1.01,0.16610560417175294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,128,power_law_1.01,0.05121920108795166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,128,power_law_1.01,0.09263359904289245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,128,power_law_1.01,0.05565440058708191
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,128,power_law_1.01,0.09575039744377137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,128,power_law_1.01,0.08466560244560242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,128,power_law_1.01,0.09705600142478943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,128,power_law_1.2,0.0611519992351532
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,128,power_law_1.01,0.145305597782135
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,128,power_law_1.01,0.024646399915218352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,128,power_law_1.01,0.10272639989852905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,128,power_law_1.2,0.0821183979511261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,128,power_law_1.01,0.1481727957725525
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,128,power_law_1.01,0.024639999866485594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,128,power_law_1.01,0.10708479881286621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,128,power_law_1.2,0.14257279634475709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,128,power_law_1.01,0.11096320152282715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,128,power_law_1.2,0.14536319971084594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,128,power_law_1.01,0.12113280296325683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,128,power_law_1.01,0.13178880214691163
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,128,power_law_1.01,0.17734400033950806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,128,power_law_1.01,0.14071680307388307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,128,power_law_1.01,0.18755199909210205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,128,power_law_1.01,0.16569600105285645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,128,power_law_1.01,0.1969151973724365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,128,balanced,0.025066666305065155
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,128,balanced,0.02740799884001414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,128,power_law_1.01,0.21810560226440429
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,128,balanced,0.02957333376010259
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,128,balanced,0.038773333032925926
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,128,balanced,0.03977066775163015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,128,power_law_1.01,0.25677440166473386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,128,balanced,0.039834665755430855
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,128,balanced,0.040021332601706185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,128,power_law_1.2,0.1451840043067932
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,128,balanced,0.05186133086681366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,128,power_law_1.01,0.2976000070571899
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,128,balanced,0.039781334499518074
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,128,balanced,0.039488000174363456
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,128,power_law_1.2,0.14813439846038817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,128,power_law_1.01,0.38161919116973875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,128,power_law_1.2,0.147052800655365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,128,power_law_1.2,0.1497215986251831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,128,power_law_1.01,0.46837759017944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,128,power_law_1.2,0.15000959634780883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,128,power_law_1.2,0.15287040472030639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,128,power_law_1.2,0.15119999647140503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,128,power_law_1.01,0.6302464008331299
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,128,power_law_1.2,0.1546239972114563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,128,power_law_1.01,0.20877439975738527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,128,power_law_1.2,0.15889279842376708
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,128,power_law_1.01,0.8096320152282714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,128,power_law_1.2,0.16410239934921264
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,128,power_law_1.01,0.02391040027141571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,128,power_law_1.2,0.16806399822235107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,128,balanced,2.5734826723734536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,128,power_law_1.01,1.160377597808838
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,128,balanced,0.03978666663169861
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,128,power_law_1.2,0.17859840393066406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,128,balanced,0.04018666595220566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,128,power_law_1.2,0.04930559992790222
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,128,power_law_1.2,0.1888576030731201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,128,power_law_1.2,0.050425601005554196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,128,power_law_1.2,0.1983296036720276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,128,power_law_1.2,0.053350400924682614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,128,power_law_1.2,0.15660159587860106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,128,power_law_1.2,0.05306879878044128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,128,power_law_1.2,0.16117119789123535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,128,power_law_1.01,0.24909439086914062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,128,power_law_1.2,0.05945600271224975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,128,power_law_1.2,0.1653887987136841
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,128,power_law_1.01,0.3379584074020386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,128,power_law_1.2,0.08844159841537476
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,128,power_law_1.01,0.0247871994972229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,128,power_law_1.2,0.09157119989395142
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,128,balanced,0.04178666571776072
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,128,power_law_1.01,0.026572799682617186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,128,power_law_1.2,0.09265919923782348
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,128,power_law_1.01,1.4601216316223145
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,128,power_law_1.01,0.02993279993534088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,128,power_law_1.2,0.09560959935188293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,128,power_law_1.2,0.2198335886001587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,128,power_law_1.01,0.03802880048751831
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,128,power_law_1.2,0.09723520278930664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,128,power_law_1.2,0.26023681163787843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,128,power_law_1.2,0.10222079753875732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,128,power_law_1.2,0.29972479343414304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,128,power_law_1.2,0.10735360383987427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,128,power_law_1.2,0.1756351947784424
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,128,power_law_1.2,0.3903872013092041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,128,power_law_1.2,0.11159679889678956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,128,power_law_1.01,0.42934398651123046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,128,power_law_1.2,0.18740479946136473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,128,power_law_1.2,0.4847104072570801
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,128,balanced,0.043893332282702126
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,128,power_law_1.2,0.19686399698257445
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,128,balanced,0.045968001087506614
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,128,balanced,0.0480320006608963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,128,power_law_1.2,0.21731839179992676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,128,balanced,0.056234667698542275
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,128,balanced,0.06458133459091187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,128,power_law_1.2,0.2572671890258789
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,128,power_law_1.01,0.03888640105724335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,128,power_law_1.2,0.2974080085754395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,128,power_law_1.01,2.1379135131835936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,128,power_law_1.01,0.03986560106277466
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,128,power_law_1.01,0.03963519930839539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,128,power_law_1.2,0.1217919945716858
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,128,power_law_1.01,0.04919039905071258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,128,power_law_1.2,0.13325439691543578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,128,power_law_1.01,0.5963776111602783
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,128,power_law_1.01,0.03919360041618347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,128,power_law_1.2,0.14248960018157958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,128,power_law_1.2,0.6735487937927246
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,128,power_law_1.01,0.03932160139083862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,128,power_law_1.01,0.7632256031036377
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,128,balanced,0.08193600177764893
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,128,power_law_1.01,0.038201600313186646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,128,power_law_1.2,0.8526975631713867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,128,power_law_1.01,0.03918080031871796
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,128,power_law_1.01,1.09683837890625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,128,power_law_1.2,0.38416640758514403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,128,power_law_1.01,0.04028159976005554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,128,power_law_1.2,1.2119872093200683
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,128,power_law_1.01,0.0421887993812561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,128,power_law_1.2,0.47730560302734376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,128,power_law_1.01,1.4345600128173828
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,128,power_law_1.01,0.045484799146652224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,128,power_law_1.2,0.6537087917327881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,128,power_law_1.01,2.8445247650146483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,128,power_law_1.2,0.16546560525894166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,128,power_law_1.2,0.8339136123657227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,128,power_law_1.01,2.0736127853393556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,128,balanced,0.10197333494822185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,128,power_law_1.2,0.20912001132965088
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,128,balanced,0.15247467160224915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,128,power_law_1.01,0.14966399669647218
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,128,balanced,0.195360004901886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,128,power_law_1.2,1.1947903633117676
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,128,balanced,0.27983466784159344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,128,power_law_1.01,0.15296000242233276
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,128,balanced,0.36133865515391034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,128,power_law_1.01,0.1540992021560669
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,128,balanced,0.033386667569478355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,128,balanced,0.05186666548252106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,128,power_law_1.01,2.775379180908203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,128,balanced,0.05000533163547516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,128,power_law_1.01,0.15920000076293944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,128,balanced,0.05117866893609365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,128,power_law_1.2,1.6140607833862304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,128,balanced,0.06850133339564006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,128,power_law_1.2,1.590982437133789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,128,balanced,0.08684800068537395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,128,balanced,0.08695466319719951
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,128,balanced,0.08873599767684937
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,128,balanced,0.09089600046475728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,128,balanced,0.09077333410580952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,128,balanced,0.09512533744176228
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,128,balanced,0.09910933176676433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,128,balanced,0.1011199951171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,128,balanced,0.1114453375339508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,128,power_law_1.2,2.3231231689453127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,128,power_law_1.2,0.25326080322265626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,128,power_law_1.2,2.3321088790893554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,128,power_law_1.2,0.348799991607666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,128,power_law_1.2,0.43943037986755373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,128,balanced,0.035690667728583016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,128,balanced,0.05077333251635233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,128,balanced,0.05216533442338308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,128,power_law_1.2,3.068825531005859
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,128,balanced,0.05110933383305868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,128,power_law_1.2,3.061203193664551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,128,balanced,0.07259733478228252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,128,balanced,0.0886293351650238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,128,power_law_1.01,0.03224959969520569
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,128,balanced,0.09090666969617207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,128,power_law_1.01,0.03375999927520752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,128,power_law_1.01,0.049958398938179015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,128,power_law_1.01,0.04938240051269531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,128,power_law_1.01,0.050809597969055174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,128,balanced,0.11735467116038005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,128,power_law_1.01,0.049344000220298764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,128,power_law_1.01,0.05084800124168396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,128,power_law_1.01,0.04973439872264862
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,128,power_law_1.2,0.023174400627613067
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,128,power_law_1.01,0.06765440106391907
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,128,power_law_1.01,0.07032960057258605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,128,power_law_1.01,0.08355200290679932
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,128,power_law_1.01,0.08671360015869141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,128,power_law_1.01,0.08538879752159119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,128,power_law_1.01,0.08843520283699036
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,128,power_law_1.01,0.04888960123062134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,128,power_law_1.01,0.08789119720458985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,128,power_law_1.2,0.608518409729004
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,128,power_law_1.01,0.060499197244644164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,128,power_law_1.01,0.08929920196533203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,128,balanced,0.09300266702969869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,128,power_law_1.01,0.09015039801597595
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,128,balanced,0.09301867087682088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,128,power_law_1.2,0.7843455791473388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,128,balanced,0.09493333101272583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,128,power_law_1.01,0.09393280148506164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,128,power_law_1.01,0.09752320051193238
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,128,balanced,0.123690664768219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,128,balanced,0.13986133535703024
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,128,power_law_1.01,0.10067199468612671
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,128,balanced,0.1718613306681315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,128,balanced,0.20316267013549805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,128,power_law_1.01,0.11000319719314575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,128,balanced,0.26503467559814453
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,128,power_law_1.2,0.0233024001121521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,128,balanced,0.3260800043741862
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,128,balanced,0.45954132080078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,128,power_law_1.2,0.02285439968109131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,128,power_law_1.01,0.1643839955329895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,128,power_law_1.01,0.09136000275611877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,128,power_law_1.2,0.023212799429893495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,128,power_law_1.01,0.16922240257263182
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,128,power_law_1.01,0.06990079879760742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,128,power_law_1.01,0.17880959510803224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,128,balanced,0.09929600358009338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,128,power_law_1.01,0.09544320106506347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,128,balanced,0.10345066587130229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,128,power_law_1.01,0.18884479999542236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,128,balanced,0.1074079970518748
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,128,power_law_1.2,0.025171199440956117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,128,balanced,0.11569066842397054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,128,power_law_1.01,0.12327040433883667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,128,power_law_1.2,1.1289600372314452
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,128,power_law_1.01,0.19804799556732178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,128,power_law_1.2,0.027884799242019653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,128,power_law_1.01,0.17043839693069457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,128,power_law_1.01,0.21973121166229248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,128,power_law_1.01,0.11726720333099365
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,128,power_law_1.2,0.03638400137424469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,128,power_law_1.01,0.21895039081573486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,128,power_law_1.2,1.4651904106140137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,128,balanced,0.5833813349405924
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,128,power_law_1.2,0.037350401282310486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,128,power_law_1.01,0.3131455898284912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,128,power_law_1.01,0.09279360175132752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,128,power_law_1.2,0.03850879967212677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,128,power_law_1.01,0.0936896026134491
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,128,power_law_1.01,0.4088319778442383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,128,power_law_1.01,0.09742720127105713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,128,power_law_1.2,2.1626943588256835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,2688,1856,6,128,1,128,power_law_1.2,0.033030399680137636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,128,power_law_1.01,0.10151040554046631
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,128,balanced,0.12201066811879475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,2688,1856,6,128,1,128,power_law_1.2,0.05322880148887634
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,128,power_law_1.01,0.10481280088424683
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,2688,1856,6,128,1,128,power_law_1.2,0.05187199711799621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,2688,1856,6,128,1,128,power_law_1.2,0.04981119930744171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,128,power_law_1.01,0.12346880435943604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,2688,1856,6,128,1,128,power_law_1.2,0.0675711989402771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,128,power_law_1.2,2.868083190917969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,2688,1856,6,128,1,128,power_law_1.2,0.08529279828071594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,128,balanced,0.8307200272878011
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,2688,1856,6,128,1,128,power_law_1.2,0.034067198634147644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,2688,1856,6,128,1,128,power_law_1.2,0.08556159734725952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,2688,1856,6,128,1,128,power_law_1.2,0.049081599712371825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,128,balanced,1.0769493579864502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,2688,1856,6,128,1,128,power_law_1.2,0.08749439716339111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,2688,1856,6,128,1,128,power_law_1.2,0.049132800102233885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,2688,1856,6,128,1,128,power_law_1.2,0.08965759873390197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,2688,1856,6,128,1,128,power_law_1.2,0.05009920001029968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,2688,1856,6,128,1,128,power_law_1.2,0.09018880128860474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,128,balanced,0.12985066572825113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,128,balanced,0.14628266294797262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,2688,1856,6,128,1,128,power_law_1.2,0.09406080245971679
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,128,balanced,0.1811573306719462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,128,power_law_1.01,0.11368319988250733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,128,balanced,0.21547732750574747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,128,power_law_1.01,0.12028160095214843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,128,power_law_1.01,0.2614207983016968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,128,power_law_1.01,0.12778879404067994
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,128,power_law_1.01,0.13976320028305053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,128,power_law_1.01,0.30103681087493894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,128,power_law_1.01,0.14433280229568482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,128,power_law_1.01,0.1725376009941101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,128,power_law_1.01,0.39007999897003176
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,128,power_law_1.01,0.2044287919998169
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,128,power_law_1.2,0.0387584000825882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,128,power_law_1.01,0.2666304111480713
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,128,power_law_1.2,0.04827519953250885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,128,power_law_1.01,0.3349503993988037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,128,balanced,1.5645227432250977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,128,power_law_1.2,0.037964800000190736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,2688,1856,6,128,1,128,power_law_1.2,0.07063680291175842
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,128,power_law_1.01,0.4838719844818115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,2688,1856,6,128,1,128,power_law_1.2,0.08727040290832519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,2688,1856,6,128,1,128,power_law_1.2,0.09811840057373047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,128,balanced,0.27794132630030316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,128,balanced,2.0644960403442383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,2688,1856,6,128,1,128,power_law_1.2,0.08850560188293458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,2688,1856,6,128,1,128,power_law_1.2,0.10121599435806275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,128,power_law_1.01,0.6275008201599122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,128,balanced,0.34651732444763184
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,2688,1856,6,128,1,128,power_law_1.2,0.09100160002708435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,2688,1856,6,128,1,128,power_law_1.2,0.11058559417724609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,128,balanced,0.48605334758758545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,2688,1856,6,128,1,128,power_law_1.2,0.09288319945335388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,2688,1856,6,128,1,128,power_law_1.2,0.11752320528030395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,128,balanced,0.6212906837463379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,128,power_law_1.01,0.9156415939331055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,2688,1856,6,128,1,128,power_law_1.2,0.09338240027427673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,2688,1856,6,128,1,128,power_law_1.2,0.12453759908676147
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,2688,1856,6,128,1,128,power_law_1.2,0.09752960205078125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,2688,1856,6,128,1,128,power_law_1.2,0.14088319540023803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,2688,1856,6,128,1,128,power_law_1.2,0.10168319940567017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,128,power_law_1.01,1.222815990447998
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,2688,1856,6,128,1,128,power_law_1.2,0.17569279670715332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,2688,1856,6,128,1,128,power_law_1.2,0.10492160320281982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,2688,1856,6,128,1,128,power_law_1.2,0.20660479068756105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,2688,1856,6,128,1,128,power_law_1.2,0.11396479606628418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,128,power_law_1.2,0.038022398948669434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,2688,1856,6,128,1,128,power_law_1.2,0.12049920558929443
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,2688,1856,6,128,1,128,power_law_1.2,0.2823040008544922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,128,power_law_1.2,0.03754880130290985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,128,power_law_1.01,1.7736255645751953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,2688,1856,6,128,1,128,power_law_1.2,0.1276479959487915
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,128,balanced,0.04357333481311798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,2688,1856,6,128,1,128,power_law_1.2,0.3753599882125854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,128,power_law_1.2,0.038553598523139956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,2688,1856,6,128,1,128,power_law_1.2,0.1431872010231018
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,128,power_law_1.2,0.039417600631713866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,2688,1856,6,128,1,128,power_law_1.2,0.5411903858184814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,128,power_law_1.01,0.17749760150909424
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,128,power_law_1.2,0.042054399847984314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,128,balanced,0.8794826666514078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,2688,1856,6,128,1,128,power_law_1.2,0.7013440132141113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,128,power_law_1.01,0.4782911777496338
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,128,power_law_1.2,0.045484799146652224
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,128,balanced,1.1380266348520915
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,128,power_law_1.2,0.04960640072822571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,128,power_law_1.01,0.6552639961242676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,2688,1856,6,128,1,128,power_law_1.2,1.0112640380859375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,128,power_law_1.2,0.061440002918243405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,128,balanced,1.6533493995666504
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,128,power_law_1.2,0.07331839799880982
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,128,power_law_1.01,0.8193344116210938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,128,balanced,0.04568000137805939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,128,balanced,0.047354668378829956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,128,balanced,0.16250133514404297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,128,balanced,2.1818453470865884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,128,balanced,0.29587199290593463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,2688,1856,6,128,1,128,power_law_1.2,0.1787392020225525
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,128,power_law_1.01,1.1650431632995606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,128,balanced,0.42084264755249023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,128,power_law_1.2,0.21250560283660888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,128,balanced,0.4291466474533081
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,2688,1856,6,128,1,128,power_law_1.01,0.20855679512023925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,128,power_law_1.01,2.4155391693115233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,128,balanced,0.43726933002471924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,128,power_law_1.2,0.28225278854370117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,128,balanced,0.06539200246334076
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,128,balanced,0.4460853338241577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,128,balanced,0.06637333333492279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,128,balanced,0.06774933139483134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,128,power_law_1.2,0.3620287895202637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,2688,1856,6,128,1,128,power_law_1.2,1.3862655639648438
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,128,balanced,0.10971200466156006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,128,balanced,0.18119466304779053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,128,balanced,0.24841066201527914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,128,power_law_1.2,0.5322495937347412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,128,balanced,0.256549338499705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,128,balanced,0.26641066869099933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,128,balanced,0.274453341960907
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,128,power_law_1.2,0.10037120580673217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,128,balanced,0.2824853261311849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,128,balanced,0.3015413284301758
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,128,power_law_1.2,0.12829439640045165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,128,balanced,0.31910934050877887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,128,balanced,0.3388426701227824
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,128,power_law_1.2,0.18413439989089966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,128,balanced,0.3753066857655843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,2688,1856,6,128,1,128,power_law_1.01,0.27239038944244387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,128,power_law_1.01,1.5244159698486328
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,128,balanced,0.4543413321177165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,2688,1856,6,128,1,128,power_law_1.01,0.340447998046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,128,balanced,0.4697386821111043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,128,balanced,0.488480011622111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,2688,1856,6,128,1,128,power_law_1.01,0.49443840980529785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,128,balanced,0.5100160042444865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,128,power_law_1.2,0.705734395980835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,128,balanced,0.5431573390960693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,2688,1856,6,128,1,128,power_law_1.01,0.646457576751709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,128,balanced,0.5770719846089681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,128,balanced,0.042768001556396484
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,128,balanced,0.04740799963474274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,128,balanced,0.04819199939568838
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,128,balanced,0.1625920037428538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,128,power_law_1.01,0.9355072021484375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,128,balanced,0.2978079915046692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,128,balanced,0.41837867101033527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,128,balanced,0.4100000063578288
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,128,balanced,0.42718935012817383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,128,balanced,0.4513333241144816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,128,power_law_1.01,1.2307071685791016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,128,balanced,0.5279733339945475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,128,balanced,0.6706826686859131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,128,balanced,0.8188266754150391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,128,power_law_1.01,2.2042688369750976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,2688,1856,6,128,1,128,power_law_1.2,1.0453696250915527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,128,balanced,1.1268906593322754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,128,balanced,0.6144373416900635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,2688,1856,6,128,1,128,power_law_1.2,1.3343680381774903
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,128,balanced,1.4152480761210124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,128,power_law_1.2,0.2425152063369751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,128,balanced,0.4376426537831624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,128,power_law_1.01,2.9276031494140624
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,128,power_law_1.2,0.3425600051879883
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,128,balanced,0.445087989171346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,128,balanced,0.45529067516326904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,128,power_law_1.01,0.08616960048675537
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,128,balanced,0.47278400262196857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,128,power_law_1.2,0.46858878135681153
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,128,balanced,0.4883626699447632
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,128,power_law_1.01,1.8811647415161132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,128,balanced,0.5108746687571207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,128,power_law_1.01,0.08630399703979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,128,power_law_1.01,0.07519360184669495
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,128,balanced,0.5452959934870402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,128,balanced,0.5834240118662516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,128,power_law_1.01,0.08669440150260925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,128,balanced,0.6850826740264893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,128,balanced,0.6228959957758585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,128,balanced,0.8249546686808268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,128,balanced,0.6967093149820963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,128,balanced,0.8388373057047526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,128,balanced,0.980725367863973
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,128,balanced,1.2745440006256104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,2688,1856,6,128,1,128,power_law_1.2,2.0328575134277345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,128,balanced,1.5514079729715984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,128,power_law_1.01,0.07559679746627808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,128,power_law_1.01,2.4215808868408204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,128,power_law_1.01,0.1640128016471863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,128,balanced,2.1431147257486978
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,128,power_law_1.01,0.0756608009338379
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,128,balanced,2.003061294555664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,128,power_law_1.01,0.0804032027721405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,128,power_law_1.01,0.16335359811782837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,2688,1856,6,128,1,128,power_law_1.2,2.6091903686523437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,128,balanced,2.706752141316732
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,128,power_law_1.01,0.08489599823951721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,128,power_law_1.01,0.2955904006958008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,128,power_law_1.2,0.0864575982093811
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,128,power_law_1.01,0.08533759713172913
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,128,power_law_1.2,0.08610560297966004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,128,power_law_1.01,0.41895041465759275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,128,power_law_1.01,0.11180800199508667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,128,power_law_1.2,0.08734719753265381
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,128,balanced,3.894293467203776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,128,power_law_1.01,0.42833919525146485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,128,balanced,0.9669280052185059
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,128,power_law_1.01,0.17870080471038818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,128,power_law_1.2,0.1624127984046936
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,128,power_law_1.01,0.2928639888763428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,128,power_law_1.01,0.43605761528015136
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,128,balanced,1.2576320171356201
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,128,power_law_1.01,0.2473151922225952
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,128,power_law_1.2,0.2922688007354736
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,128,power_law_1.01,0.44495358467102053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,128,power_law_1.01,0.4159552097320557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,128,power_law_1.01,0.2563711881637573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,128,power_law_1.01,0.4539968013763428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,128,balanced,5.071776072184245
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,128,power_law_1.2,0.41667838096618653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,128,power_law_1.01,0.42556161880493165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,128,power_law_1.01,0.2622208118438721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,128,power_law_1.01,0.47205119132995604
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,128,power_law_1.2,0.42627201080322263
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,128,power_law_1.01,0.4334271907806396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,128,power_law_1.01,0.2738368034362793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,128,power_law_1.01,0.49039359092712403
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,128,power_law_1.01,0.4413440227508545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,128,power_law_1.01,0.2823935985565186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,128,power_law_1.01,0.5115200042724609
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,128,power_law_1.01,0.45032320022583006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,128,power_law_1.01,0.300927996635437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,128,balanced,2.571077346801758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,128,power_law_1.01,0.315775990486145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,128,balanced,1.5376267433166504
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,128,balanced,7.4732106526692705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,128,power_law_1.01,0.33955199718475343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,128,balanced,2.1232959429423013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,128,power_law_1.01,0.3738816022872925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,128,balanced,3.7416906356811523
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,128,power_law_1.2,0.43224320411682127
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,128,balanced,2.6957225799560547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,128,power_law_1.2,0.443071985244751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,128,power_law_1.01,0.4672383785247803
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,128,power_law_1.2,0.4507904052734375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,128,power_law_1.01,0.48492798805236814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,128,balanced,4.902096112569173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,128,balanced,3.8835627237955728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,128,power_law_1.01,0.5048255920410156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,128,balanced,9.806400299072266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,128,power_law_1.01,0.5394239902496338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,128,power_law_1.01,0.4166399955749512
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,128,power_law_1.2,0.07576320171356202
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,128,power_law_1.01,0.46460800170898436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,128,power_law_1.01,0.5463488101959229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,2688,1856,6,128,1,128,power_law_1.2,2.0894079208374023
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,128,balanced,5.028490702311198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,128,power_law_1.01,0.5451839923858642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,128,power_law_1.2,0.467468786239624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,128,balanced,7.246682484944661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,128,power_law_1.01,0.7014143943786622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,128,power_law_1.2,0.4855040073394775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,128,power_law_1.01,0.8574080467224121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,128,power_law_1.2,0.5066112041473388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,128,power_law_1.2,0.5441279888153077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,128,power_law_1.01,1.1758720397949218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,128,power_law_1.2,0.07614079713821412
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,128,balanced,7.411877314249675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,128,power_law_1.2,0.579366397857666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,128,power_law_1.01,0.5853312015533447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,128,power_law_1.2,0.07561600208282471
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,128,power_law_1.2,0.6177599906921387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,128,power_law_1.01,0.6269375801086425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,128,power_law_1.2,0.16472320556640624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,128,balanced,9.55287996927897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,128,power_law_1.2,0.6962944030761719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,128,power_law_1.01,0.7039680004119873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,128,power_law_1.2,0.29431040287017823
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,128,power_law_1.2,0.07949439883232116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,128,power_law_1.01,0.5756095886230469
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,128,power_law_1.2,0.8487423896789551
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,128,power_law_1.01,0.8809599876403809
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,128,power_law_1.2,0.4176767826080322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,128,power_law_1.2,0.08424959778785705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,128,power_law_1.01,0.6170303821563721
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,128,power_law_1.2,1.047424030303955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,128,power_law_1.01,1.4849791526794434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,128,power_law_1.2,0.08451840281486511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,128,power_law_1.01,0.6880320072174072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,128,balanced,9.72382926940918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,256,balanced,0.0658079981803894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,128,power_law_1.2,0.10577919483184814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,128,power_law_1.01,0.8423423767089844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,128,power_law_1.01,2.103001594543457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,128,power_law_1.2,0.17624959945678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,128,power_law_1.01,1.0054207801818849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,128,power_law_1.2,0.24591999053955077
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,128,power_law_1.01,1.341971206665039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,128,power_law_1.2,0.2559040069580078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,128,power_law_1.01,1.0248127937316895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,128,power_law_1.01,2.71844482421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,128,power_law_1.2,0.26248960494995116
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,256,balanced,0.1308746635913849
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,128,power_law_1.2,0.2734143972396851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,128,power_law_1.01,1.645439910888672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,128,power_law_1.2,1.3887295722961426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,128,power_law_1.2,0.2825344085693359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,128,power_law_1.2,0.29869439601898196
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,128,power_law_1.2,1.724403190612793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,128,power_law_1.01,2.265900802612305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,128,power_law_1.2,0.31651198863983154
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,128,power_law_1.01,3.9455936431884764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,128,power_law_1.2,0.42590718269348143
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,128,power_law_1.2,0.3373183965682983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,2688,1856,6,128,1,128,power_law_1.2,2.6698368072509764
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,256,balanced,0.046207999189694725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,256,balanced,0.08067733546098073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,256,balanced,0.08136000235875447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,128,power_law_1.2,0.3805567979812622
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,256,balanced,0.09913600484530131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,128,power_law_1.01,1.3774463653564453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,128,power_law_1.01,2.942374420166016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,128,power_law_1.2,0.42031359672546387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,128,power_law_1.2,0.46574082374572756
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,128,power_law_1.01,1.698214340209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,128,power_law_1.01,5.20909423828125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,128,power_law_1.2,0.551251220703125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,128,power_law_1.01,2.3421184539794924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,128,power_law_1.2,2.356505584716797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,256,balanced,0.13809067010879517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,256,balanced,0.24995734294255575
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,256,balanced,0.25473066171010333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,256,balanced,0.25667200485865277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,256,balanced,0.2610879937807719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,256,balanced,0.26501333713531494
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,256,balanced,0.27542932828267414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,256,balanced,0.2818453311920166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,128,power_law_1.2,3.1343231201171875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,256,balanced,0.28969599803288776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,256,balanced,0.3080586592356364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,256,balanced,0.3240533272425334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,256,balanced,0.3421440124511719
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,128,power_law_1.2,0.7093632221221924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,128,power_law_1.2,0.43535361289978025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,128,power_law_1.01,7.734636688232422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,256,balanced,0.1322773297627767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,256,balanced,0.14046399792035422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,256,balanced,0.1807360053062439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,256,balanced,0.2527199983596802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,128,power_law_1.2,0.8792511940002441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,256,balanced,0.2605813344319661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,256,balanced,0.26497600475947064
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,256,balanced,0.2725759943326314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,128,power_law_1.2,4.571270370483399
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,256,balanced,0.2754506667455037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,128,power_law_1.01,2.990732765197754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,256,balanced,0.3749973376592
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,256,balanced,0.4422773520151774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,256,balanced,0.5106720129648844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,128,power_law_1.2,0.4439871788024902
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,256,balanced,0.6395519971847534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,128,power_law_1.2,0.45272321701049806
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,128,power_law_1.01,4.303180694580078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,128,power_law_1.2,6.044710540771485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,128,power_law_1.2,1.1927616119384765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,128,power_law_1.01,10.262822723388672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,128,power_law_1.01,4.162284851074219
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,256,balanced,0.07422400017579396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,256,balanced,0.0730560024579366
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,256,balanced,0.08026133477687836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,256,balanced,0.07940799991289775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,256,balanced,0.09404266873995464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,256,balanced,0.14317333698272705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,256,balanced,0.14686933159828186
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,256,balanced,0.15109866857528687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,256,balanced,0.1545973320802053
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,256,balanced,0.7689440250396729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,256,balanced,0.1576746702194214
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,128,power_law_1.01,5.598931121826172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,256,balanced,0.1661066710948944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,128,power_law_1.2,0.4691648006439209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,256,balanced,0.28351465861002606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,256,balanced,0.17486933867136636
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,256,balanced,1.0276426474253337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,256,balanced,0.29063467184702557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,256,balanced,0.2964213291803996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,128,power_law_1.2,0.48691840171813966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,256,balanced,0.31014400720596313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,256,balanced,1.2837226390838623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,256,balanced,0.3244640032450358
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,128,power_law_1.2,0.50862717628479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,256,balanced,0.33936532338460285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,128,power_law_1.01,5.4970752716064455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,256,balanced,0.37034134070078534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,256,balanced,1.8012852668762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,128,power_law_1.2,9.177542114257813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,128,power_law_1.2,0.5457727909088135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,128,power_law_1.2,0.5854080200195313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,256,balanced,2.316640059153239
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,256,balanced,0.18300267060597739
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,256,balanced,0.19787200291951498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,256,balanced,0.2136426568031311
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,256,balanced,0.22987200816472372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,256,balanced,0.25913600126902264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,128,power_law_1.2,1.5084992408752442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,256,balanced,3.351290702819824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,256,balanced,0.32531734307607013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,256,balanced,0.38654931386311847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,256,balanced,0.43575998147328693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,256,balanced,0.5075413386027018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,256,balanced,0.49941333134969074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,256,balanced,0.6265813509623209
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,256,balanced,0.6205759843190511
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,128,power_law_1.2,2.1607872009277345
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,128,power_law_1.01,8.171084594726562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,256,balanced,0.7440213362375895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,256,balanced,4.387653350830078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,128,power_law_1.2,0.6262144088745117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,256,balanced,0.9907893339792887
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,256,balanced,1.2321013609568279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,128,power_law_1.01,8.189600372314453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,128,power_law_1.2,2.8050432205200195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,256,balanced,1.7271092732747395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,256,balanced,0.8687307039896647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,256,balanced,2.218671957651774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,256,balanced,1.108954668045044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,128,power_law_1.2,4.224319839477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,128,power_law_1.2,0.7095424175262451
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,256,balanced,1.587546666463216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,256,balanced,3.2123947143554688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,128,power_law_1.2,0.8866880416870118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,256,balanced,2.06987730662028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,128,power_law_1.01,11.018080139160157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,128,power_law_1.2,1.0556608200073243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,256,power_law_1.01,0.061337602138519284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,256,balanced,3.033578554789225
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,256,power_law_1.01,0.12410240173339844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,128,power_law_1.2,5.317580795288086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,128,power_law_1.01,10.959661102294922
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,128,power_law_1.2,1.397158432006836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,256,power_law_1.01,0.12588800191879274
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,256,power_law_1.01,0.04757120013237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,256,power_law_1.01,0.127347195148468
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,256,power_law_1.01,0.07738239765167236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,256,power_law_1.01,0.17314560413360597
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,128,power_law_1.2,1.759654426574707
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,256,balanced,0.02956799914439519
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,256,power_law_1.01,0.08007680177688599
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,256,balanced,0.029525332152843475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,256,power_law_1.01,0.25026559829711914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,256,balanced,4.204048156738281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,256,power_law_1.01,0.0981760025024414
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,256,power_law_1.01,0.26131839752197267
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,256,power_law_1.01,0.06904320120811462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,256,power_law_1.01,0.2637376070022583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,128,power_law_1.2,2.4522560119628904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,256,power_law_1.01,0.07310720086097718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,256,power_law_1.01,0.26892800331115724
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,256,power_law_1.01,0.07885439991950989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,256,power_law_1.01,0.273471999168396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,256,power_law_1.01,0.0807424008846283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,256,balanced,3.9994080861409507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,128,power_law_1.2,8.128550720214843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,256,power_law_1.01,0.09910399913787842
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,256,balanced,0.029535998900731403
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,256,balanced,0.03159466634194056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,256,power_law_1.01,0.14186240434646608
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,256,balanced,0.03226666649182638
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,256,balanced,0.04381866753101349
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,256,balanced,0.06032533446947733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,256,power_law_1.01,0.13719040155410767
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,256,balanced,0.04590400060017904
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,256,balanced,0.061797335743904114
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,256,balanced,0.06227200229962667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,256,power_law_1.01,0.24899840354919434
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,256,balanced,0.04771733283996582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,256,balanced,0.04798933366934458
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,256,balanced,0.048698668678601585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,256,power_law_1.01,0.25290238857269287
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,256,balanced,0.05026133358478546
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,256,balanced,0.05014933149019877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,256,power_law_1.01,0.2563456058502197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,256,power_law_1.01,0.2809344053268433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,128,power_law_1.2,11.960582733154297
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,256,power_law_1.01,0.26087040901184083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,128,power_law_1.2,3.1393856048583983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,7168,2048,8,256,1,256,power_law_1.2,0.07535359859466553
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,256,power_law_1.01,0.26439681053161623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,7168,2048,8,256,1,256,power_law_1.2,0.12993279695510865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,256,power_law_1.01,0.14650239944458007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,256,power_law_1.01,0.27326719760894774
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,7168,2048,8,256,1,256,power_law_1.2,0.1314239978790283
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,256,power_law_1.01,0.1486143946647644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,256,power_law_1.01,0.2814719915390015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,7168,2048,8,256,1,256,power_law_1.2,0.1351680040359497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,256,power_law_1.01,0.15259519815444947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,256,power_law_1.01,0.29171841144561766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,7168,2048,8,256,1,256,power_law_1.2,0.16895999908447265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,256,power_law_1.01,0.1579327940940857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,128,power_law_1.2,11.05038070678711
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,256,balanced,0.05310933291912079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,128,power_law_1.2,4.518783950805664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,7168,2048,8,256,1,256,power_law_1.2,0.2488703966140747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,7168,2048,8,256,1,256,power_law_1.2,0.047295999526977536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,256,power_law_1.01,0.28827519416809083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,7168,2048,8,256,1,256,power_law_1.2,0.25717759132385254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,7168,2048,8,256,1,256,power_law_1.2,0.07579519748687744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,256,power_law_1.01,0.29543681144714357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,7168,2048,8,256,1,256,power_law_1.2,0.26359679698944094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,7168,2048,8,256,1,256,power_law_1.2,0.07850880026817322
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,256,power_law_1.01,0.30743680000305174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,7168,2048,8,256,1,256,power_law_1.2,0.2670016050338745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,7168,2048,8,256,1,256,power_law_1.2,0.09743360280990601
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,256,power_law_1.01,0.3217407941818237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,7168,2048,8,256,1,256,power_law_1.2,0.2725311994552612
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,256,power_law_1.01,0.028262400627136232
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,7168,2048,8,256,1,256,power_law_1.2,0.13672959804534912
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,256,power_law_1.01,0.33755519390106203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,7168,2048,8,256,1,256,power_law_1.2,0.28086400032043457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,256,power_law_1.01,0.028275200724601747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,256,power_law_1.01,0.3688960075378418
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,256,balanced,0.056757330894470215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,256,power_law_1.01,0.16753920316696166
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,256,balanced,0.06329066554705302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,128,power_law_1.2,6.089152145385742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,256,power_law_1.01,0.435756778717041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,256,power_law_1.01,0.17440639734268187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,256,power_law_1.01,0.18179839849472046
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,256,power_law_1.01,0.5005631923675538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,256,power_law_1.01,0.19530240297317505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,256,power_law_1.01,0.6263999938964844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,256,power_law_1.01,0.2122431993484497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,256,power_law_1.01,0.22992000579833985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,256,power_law_1.01,0.7570816040039062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,256,power_law_1.01,0.3076159954071045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,7168,2048,8,256,1,256,power_law_1.2,0.24863998889923095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,7168,2048,8,256,1,256,power_law_1.2,0.28828160762786864
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,256,power_law_1.01,1.0433728218078613
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,7168,2048,8,256,1,256,power_law_1.2,0.2526015996932983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,7168,2048,8,256,1,256,power_law_1.2,0.2955967903137207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,7168,2048,8,256,1,256,power_law_1.2,0.2551935911178589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,256,power_law_1.01,1.323852825164795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,7168,2048,8,256,1,256,power_law_1.2,0.2592128038406372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,7168,2048,8,256,1,256,power_law_1.2,0.2642496109008789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,256,power_law_1.01,1.9298751831054688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,7168,2048,8,256,1,256,power_law_1.2,0.27232000827789304
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,256,power_law_1.01,0.2629055976867676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,7168,2048,8,256,1,256,power_law_1.2,0.2809344053268433
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,256,power_law_1.01,0.32468481063842775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,256,power_law_1.01,0.3326272010803223
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,256,balanced,0.0708000014225642
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,7168,2048,8,256,1,256,power_law_1.2,0.290720009803772
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,256,balanced,0.08950400352478027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,256,power_law_1.01,0.34328958988189695
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,256,power_law_1.01,0.39820160865783694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,7168,2048,8,256,1,256,power_law_1.2,0.3082175970077515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,256,power_law_1.01,0.3767359972000122
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,256,power_law_1.01,0.5263423919677734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,7168,2048,8,256,1,256,power_law_1.2,0.32269439697265623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,256,power_law_1.01,0.4479231834411621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,7168,2048,8,256,1,256,power_law_1.2,0.3384000062942505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,256,power_law_1.01,0.6626688003540039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,256,power_law_1.01,0.5161983966827393
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,7168,2048,8,256,1,256,power_law_1.2,0.37093119621276854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,256,power_law_1.01,0.9210687637329101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,7168,2048,8,256,1,256,power_law_1.2,0.439788818359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,256,power_law_1.01,0.6588992118835449
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,256,power_law_1.01,0.028415998816490172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,7168,2048,8,256,1,256,power_law_1.2,0.5080319881439209
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,256,power_law_1.01,0.03049600124359131
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,256,power_law_1.01,0.8032511711120606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,256,power_law_1.01,1.190995216369629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,256,power_law_1.01,2.5145727157592774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,256,power_law_1.01,0.031916800141334536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,7168,2048,8,256,1,256,power_law_1.2,0.6519680023193359
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,256,power_law_1.01,1.1016960144042969
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,256,power_law_1.01,0.041715198755264284
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,256,power_law_1.01,1.713158416748047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,7168,2048,8,256,1,256,power_law_1.2,0.8064703941345215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,256,power_law_1.01,0.055052798986434934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,256,power_law_1.01,1.4009023666381837
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,256,power_law_1.01,0.04410879909992218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,7168,2048,8,256,1,256,power_law_1.2,1.1387200355529785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,256,power_law_1.01,3.647200012207031
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,256,power_law_1.01,0.05663999915122986
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,256,power_law_1.01,2.265945625305176
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,256,power_law_1.01,0.055936002731323244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,7168,2048,8,256,1,256,power_law_1.2,0.3075455904006958
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,7168,2048,8,256,1,256,power_law_1.2,1.4448384284973144
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,256,power_law_1.01,0.04444159865379334
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,7168,2048,8,256,1,256,power_law_1.2,0.32531840801239015
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,256,power_law_1.01,0.04535039961338043
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,7168,2048,8,256,1,256,power_law_1.2,0.3416703939437866
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,256,power_law_1.01,0.04723840057849884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,7168,2048,8,256,1,256,power_law_1.2,2.1595008850097654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,7168,2048,8,256,1,256,power_law_1.2,0.3775167942047119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,256,power_law_1.01,3.3273216247558595
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,256,power_law_1.01,0.04750719964504242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,256,power_law_1.01,4.882803344726563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,7168,2048,8,256,1,256,power_law_1.2,0.44983677864074706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,256,power_law_1.01,0.049593600630760196
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,256,power_law_1.01,0.05122560262680054
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,256,balanced,0.10965866843859355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,256,power_law_1.01,2.029734420776367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,256,power_law_1.01,0.05538560152053833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,7168,2048,8,256,1,256,power_law_1.2,2.904742431640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,256,power_law_1.01,0.06424319744110107
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,256,power_law_1.01,0.07114239931106567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,256,power_law_1.01,4.3943359375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,256,power_law_1.01,0.09675520062446594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,256,power_law_1.01,2.624518394470215
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,256,power_law_1.01,0.1214784026145935
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,256,power_law_1.01,0.1700096011161804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,7168,2048,8,256,1,256,power_law_1.2,4.1389312744140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,7168,2048,8,256,1,256,power_law_1.2,0.0732159972190857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,256,power_law_1.01,0.21928958892822265
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,256,balanced,0.1535253326098124
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,256,balanced,0.19542932510375977
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,256,power_law_1.01,0.31510400772094727
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,256,balanced,0.28519999980926514
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,256,power_law_1.01,3.8229183197021483
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,256,balanced,0.3683040142059326
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,256,power_law_1.01,0.4149631977081299
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,256,balanced,0.5247040192286173
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,7168,2048,8,256,1,256,power_law_1.2,5.9680320739746096
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,7168,2048,8,256,1,256,power_law_1.2,0.5224575996398926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,7168,2048,8,256,1,256,power_law_1.2,0.07048959732055664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,7168,2048,8,256,1,256,power_law_1.2,0.0763264000415802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,256,balanced,0.04876266419887543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,256,balanced,0.6870986620585123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,256,balanced,0.09924266735712688
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,256,balanced,0.10030933221181233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,256,balanced,0.05005866785844167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,256,power_law_1.01,0.5877888202667236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,256,balanced,0.10941867033640544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,256,balanced,0.07396266857783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,256,balanced,0.13778666655222574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,256,balanced,0.07242133220036824
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,256,balanced,0.0879146655400594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,256,power_law_1.01,5.019295883178711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,256,balanced,0.22779732942581177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,7168,2048,8,256,1,256,power_law_1.2,0.6691967964172363
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,256,balanced,0.1279146671295166
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,256,balanced,0.232314666112264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,256,balanced,0.22822932402292886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,256,balanced,0.23509865999221802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,256,balanced,0.23224000136057535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,256,balanced,0.23818665742874146
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,256,balanced,0.2346186637878418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,256,balanced,0.2422773241996765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,256,balanced,0.24041599035263062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,256,balanced,0.2485599915186564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,256,balanced,0.24169067541758218
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,128,power_law_1.2,9.513452911376953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,256,balanced,0.2589120070139567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,256,balanced,0.24965333938598633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,256,balanced,0.26608532667160034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,256,balanced,0.28008000055948895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,256,balanced,0.2936319907506307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,256,balanced,0.30777599414189655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,256,balanced,0.3386933406194051
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,256,balanced,0.3976586659749349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,256,balanced,0.45668800671895343
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,256,balanced,0.5718346834182739
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,7168,2048,8,256,1,256,power_law_1.2,0.026752001047134398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,256,power_law_1.01,0.7863679885864258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,256,balanced,0.682645320892334
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,7168,2048,8,256,1,256,power_law_1.2,0.02717440128326416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,7168,2048,8,256,1,256,power_law_1.2,0.8403008460998536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,7168,2048,8,256,1,256,power_law_1.2,0.07768959999084472
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,7168,2048,8,256,1,256,power_law_1.2,0.02698880136013031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,256,balanced,0.2588213284810384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,7168,2048,8,256,1,256,power_law_1.2,0.09690240025520325
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,256,balanced,0.26499199867248535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,7168,2048,8,256,1,256,power_law_1.2,0.02890239953994751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,7168,2048,8,256,1,256,power_law_1.2,1.178662395477295
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,256,balanced,0.2795520027478536
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,256,balanced,0.29681599140167236
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,7168,2048,8,256,1,256,power_law_1.2,0.030969598889350893
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,7168,2048,8,256,1,256,power_law_1.2,0.041817599534988405
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,7168,2048,8,256,1,256,power_law_1.2,0.06360960006713867
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,7168,2048,8,256,1,256,power_law_1.2,0.04411520063877106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,256,balanced,0.910912036895752
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,7168,2048,8,256,1,256,power_law_1.2,0.061862397193908694
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,256,balanced,1.136677344640096
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,7168,2048,8,256,1,256,power_law_1.2,0.061260801553726194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,7168,2048,8,256,1,256,power_law_1.2,0.14201600551605226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,256,balanced,1.5899839401245117
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,7168,2048,8,256,1,256,power_law_1.2,1.5228416442871093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,7168,2048,8,256,1,256,power_law_1.2,0.1460736036300659
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,256,balanced,2.0422239303588867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,7168,2048,8,256,1,256,power_law_1.2,0.14920320510864257
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,7168,2048,8,256,1,256,power_law_1.2,0.15212160348892212
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,7168,2048,8,256,1,256,power_law_1.2,2.273651123046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,256,balanced,2.9599040349324546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,7168,2048,8,256,1,256,power_law_1.2,0.15711359977722167
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,256,balanced,0.029391999046007793
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,256,balanced,0.028362666567166645
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,256,balanced,0.029482667644818623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,256,balanced,0.3131360014279683
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,7168,2048,8,256,1,256,power_law_1.2,0.043731200695037845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,256,balanced,0.34275734424591064
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,7168,2048,8,256,1,256,power_law_1.2,0.04417920112609863
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,7168,2048,8,256,1,256,power_law_1.2,2.966713523864746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,256,balanced,3.887903849283854
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,7168,2048,8,256,1,256,power_law_1.2,0.04643200039863586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,256,power_law_1.01,0.04686079919338226
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,7168,2048,8,256,1,256,power_law_1.2,0.046291199326515195
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,256,power_law_1.01,0.09804159998893738
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,256,power_law_1.01,0.09985920190811157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,256,power_law_1.01,0.1088703989982605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,7168,2048,8,256,1,256,power_law_1.2,0.16499840021133422
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,7168,2048,8,256,1,256,power_law_1.2,4.254489517211914
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,256,balanced,0.4057493209838867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,7168,2048,8,256,1,256,power_law_1.2,0.1720703959465027
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,256,balanced,0.46898667017618817
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,7168,2048,8,256,1,256,power_law_1.2,0.18165119886398315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,256,balanced,0.5911306540171305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,7168,2048,8,256,1,256,power_law_1.2,0.1972864031791687
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,7168,2048,8,256,1,256,power_law_1.2,0.04878079891204834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,7168,2048,8,256,1,256,power_law_1.2,0.2167423963546753
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,7168,2048,8,256,1,256,power_law_1.2,0.05008000135421753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,7168,2048,8,256,1,256,power_law_1.2,0.23075199127197266
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,7168,2048,8,256,1,256,power_law_1.2,0.055199998617172244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,7168,2048,8,256,1,256,power_law_1.2,0.26721279621124266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,256,power_law_1.01,0.13740160465240478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,7168,2048,8,256,1,256,power_law_1.2,5.938719940185547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,7168,2048,8,256,1,256,power_law_1.2,0.3388927936553955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,256,power_law_1.01,0.22588160037994384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,256,power_law_1.01,0.04659200012683869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,7168,2048,8,256,1,256,power_law_1.2,0.41390719413757326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,256,balanced,0.7097600301106771
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,256,power_law_1.01,0.07054719924926758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,7168,2048,8,256,1,256,power_law_1.2,0.5479360103607178
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,256,balanced,0.9491573174794515
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,256,power_law_1.01,0.07118719816207886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,256,balanced,1.187007983525594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,7168,2048,8,256,1,256,power_law_1.2,0.6883008003234863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,256,balanced,0.030810666580994923
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,256,balanced,0.0336053321758906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,256,balanced,1.6668960253397624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,7168,2048,8,256,1,256,power_law_1.2,0.9620287895202637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,256,power_law_1.01,0.23054718971252441
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,7168,2048,8,256,1,256,power_law_1.2,1.254035186767578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,256,power_law_1.01,0.23306879997253419
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,256,power_law_1.01,0.23790080547332765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,256,power_law_1.01,0.08651520013809204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,256,power_law_1.01,0.24120960235595704
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,7168,2048,8,256,1,256,power_law_1.2,0.06251519918441772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,7168,2048,8,256,1,256,power_law_1.2,1.825119972229004
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,256,power_law_1.01,0.12557439804077147
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,256,balanced,0.04176533222198486
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,7168,2048,8,256,1,256,power_law_1.2,0.07395840287208558
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,256,balanced,0.04275733232498169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,256,power_law_1.01,0.22746880054473878
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,256,balanced,0.0561706672112147
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,256,balanced,0.055776000022888184
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,7168,2048,8,256,1,256,power_law_1.2,0.10341119766235352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,256,power_law_1.01,0.23147521018981934
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,256,balanced,0.06663466493288676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,256,balanced,0.06608533362547557
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,7168,2048,8,256,1,256,power_law_1.2,0.12872320413589478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,256,power_law_1.01,0.23395841121673583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,7168,2048,8,256,1,256,power_law_1.2,2.4746496200561525
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,7168,2048,8,256,1,256,power_law_1.2,0.18078080415725709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,256,power_law_1.01,0.23812479972839357
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,7168,2048,8,256,1,256,power_law_1.2,0.23772799968719482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,256,power_law_1.01,0.24090240001678467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,256,power_law_1.01,0.24897279739379882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,7168,2048,8,256,1,256,power_law_1.2,0.3461375951766968
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,256,power_law_1.01,0.2571007966995239
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,7168,2048,8,256,1,256,power_law_1.2,0.4799039840698242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,7168,2048,8,256,1,256,power_law_1.2,3.686265563964844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,256,power_law_1.01,0.2641279935836792
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,256,balanced,0.05644266804059347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,256,balanced,2.1468960444132485
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,256,balanced,0.043866669138272606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,256,power_law_1.01,0.2776639938354492
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,7168,2048,8,256,1,256,power_law_1.2,0.6867392063140869
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,256,power_law_1.01,0.2929088115692139
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,256,power_law_1.01,0.30799360275268556
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,7168,2048,8,256,1,256,power_law_1.2,0.9181183815002442
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,256,balanced,3.105818748474121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,256,power_law_1.01,0.3381632089614868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,256,power_law_1.01,0.06538239717483521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,256,power_law_1.01,0.24912641048431397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,256,power_law_1.01,0.4005760192871094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,256,power_law_1.01,0.06787840127944947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,7168,2048,8,256,1,256,power_law_1.2,4.931808090209961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,256,power_law_1.01,0.46096000671386717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,256,power_law_1.01,0.07129600048065185
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,256,balanced,4.073557217915853
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,256,power_law_1.01,0.5806272029876709
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,256,balanced,0.07548800110816956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,256,balanced,0.0739519993464152
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,256,balanced,0.08928533395131429
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,256,balanced,0.1317813297112783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,256,power_law_1.01,0.7002367973327637
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,256,power_law_1.01,0.967961597442627
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,256,power_law_1.01,0.256499195098877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,256,power_law_1.01,1.2347392082214355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,256,power_law_1.01,0.26344320774078367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,256,power_law_1.01,0.07169920206069946
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,256,power_law_1.01,0.27956480979919435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,6144,2048,8,256,1,256,power_law_1.2,0.044838398694992065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,256,power_law_1.01,1.7441600799560546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,256,power_law_1.01,0.08780800104141236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,256,power_law_1.01,0.2959872007369995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,6144,2048,8,256,1,256,power_law_1.2,0.0880832016468048
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,256,power_law_1.01,0.1307584047317505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,6144,2048,8,256,1,256,power_law_1.2,0.08824319839477539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,256,power_law_1.01,0.1343168020248413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,6144,2048,8,256,1,256,power_law_1.2,0.10529279708862305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,256,power_law_1.01,2.2852863311767577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,256,power_law_1.01,0.13733760118484498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,6144,2048,8,256,1,256,power_law_1.2,0.14481279850006104
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,256,power_law_1.01,0.14074239730834961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,6144,2048,8,256,1,256,power_law_1.2,0.22544000148773194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,256,power_law_1.01,0.14306559562683105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,6144,2048,8,256,1,256,power_law_1.2,0.2304447889328003
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,256,balanced,0.13598400354385376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,256,balanced,0.13809067010879517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,256,power_law_1.01,3.3298110961914062
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,256,power_law_1.01,0.027167999744415285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,6144,2048,8,256,1,256,power_law_1.2,0.2333695888519287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,256,power_law_1.01,0.31055359840393065
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,6144,2048,8,256,1,256,power_law_1.2,0.2369152069091797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,256,power_law_1.01,0.3397056102752686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,6144,2048,8,256,1,256,power_law_1.2,0.2404223918914795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,256,power_law_1.01,0.39828479290008545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,6144,2048,8,256,1,256,power_law_1.2,0.24908161163330078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,256,power_law_1.01,0.4620863914489746
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,6144,2048,8,256,1,256,power_law_1.2,0.25708799362182616
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,256,balanced,0.04568533102671305
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,256,balanced,0.04624533156553904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,6144,2048,8,256,1,256,power_law_1.2,0.2639616012573242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,256,power_law_1.01,4.399027252197266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,256,power_law_1.01,0.5822015762329101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,256,power_law_1.01,0.15123840570449829
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,6144,2048,8,256,1,256,power_law_1.2,0.27762560844421386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,256,power_law_1.01,0.15890560150146485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,256,balanced,0.1439413328965505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,256,power_law_1.01,0.7085184097290039
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,6144,2048,8,256,1,256,power_law_1.2,0.29421439170837405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,256,power_law_1.01,0.16603519916534423
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,256,power_law_1.01,0.02693760097026825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,6144,2048,8,256,1,256,power_law_1.2,0.3087104082107544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,256,power_law_1.01,0.1797760009765625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,256,power_law_1.01,0.9827136039733887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,256,power_law_1.01,0.027641600370407103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,6144,2048,8,256,1,256,power_law_1.2,0.33942399024963377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,256,power_law_1.01,0.19555840492248536
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,256,power_law_1.01,0.029363200068473816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,256,power_law_1.01,0.21298561096191407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,6144,2048,8,256,1,256,power_law_1.2,0.40448641777038574
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,256,power_law_1.01,1.2338624000549316
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,128,power_law_1.2,12.462438201904297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,256,power_law_1.01,0.031001600623130798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,6144,2048,8,256,1,256,power_law_1.2,0.46561279296875
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,256,balanced,0.04752533137798309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,6144,2048,8,256,1,256,power_law_1.2,0.06596480011940002
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,256,balanced,0.04846400022506714
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,6144,2048,8,256,1,256,power_law_1.2,0.04659200012683869
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,256,balanced,0.05002133548259735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,6144,2048,8,256,1,256,power_law_1.2,0.5953023910522461
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,256,balanced,0.05234666665395101
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,6144,2048,8,256,1,256,power_law_1.2,0.06833279728889466
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,256,balanced,0.05843733251094818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,256,power_law_1.01,1.776153564453125
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,256,balanced,0.06434666613737743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,6144,2048,8,256,1,256,power_law_1.2,0.07267839908599853
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,256,balanced,0.0805920014778773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,6144,2048,8,256,1,256,power_law_1.2,0.7516543865203857
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,256,balanced,0.09769599636395772
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,6144,2048,8,256,1,256,power_law_1.2,0.06978560090065003
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,256,balanced,0.13838932911554971
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,256,balanced,0.17510932683944702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,6144,2048,8,256,1,256,power_law_1.2,1.0680831909179687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,6144,2048,8,256,1,256,power_law_1.2,0.08476160168647766
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,256,balanced,0.25470399856567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,256,power_law_1.01,2.328294372558594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,256,balanced,0.3280959924062093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,6144,2048,8,256,1,256,power_law_1.2,0.13080320358276368
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,256,balanced,0.4652213255564372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,6144,2048,8,256,1,256,power_law_1.2,1.3390463829040526
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,6144,2048,8,256,1,256,power_law_1.2,0.13434879779815673
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,256,power_law_1.01,0.04007680118083954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,256,balanced,0.607264002164205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,256,balanced,0.14614400267601013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,6144,2048,8,256,1,256,power_law_1.2,0.1365504026412964
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,256,power_law_1.01,0.04162560105323791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,256,balanced,0.15338666240374246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,256,balanced,0.1611253321170807
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,256,balanced,0.16885334253311157
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,6144,2048,8,256,1,256,power_law_1.2,0.14065279960632324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,256,power_law_1.01,0.053420799970626834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,6144,2048,8,256,1,256,power_law_1.2,1.9588031768798828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,256,balanced,0.18205867211023966
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,256,power_law_1.01,3.4224830627441407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,256,balanced,0.19753066698710123
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,6144,2048,8,256,1,256,power_law_1.2,0.1431872010231018
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,256,balanced,0.20990933974583945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,256,balanced,0.2389706571896871
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,6144,2048,8,256,1,256,power_law_1.2,0.1522112011909485
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,256,balanced,0.2988426685333252
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,256,balanced,0.36035199960072833
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,6144,2048,8,256,1,256,power_law_1.2,0.1587839961051941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,256,power_law_1.01,0.24250879287719726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,256,balanced,0.4738453229268392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,6144,2048,8,256,1,256,power_law_1.2,2.579635238647461
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,6144,2048,8,256,1,256,power_law_1.2,0.16582399606704712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,256,balanced,0.5849813222885132
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,256,power_law_1.01,0.3064448118209839
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,6144,2048,8,256,1,256,power_law_1.2,0.18039679527282715
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,256,balanced,0.8152639865875244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,256,power_law_1.01,0.37043840885162355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,6144,2048,8,256,1,256,power_law_1.2,0.07103999853134155
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,6144,2048,8,256,1,256,power_law_1.2,0.19865599870681763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,256,power_law_1.01,4.465119934082031
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,256,balanced,0.031386665999889374
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,256,power_law_1.01,0.49290242195129397
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,256,balanced,0.052416001756985985
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,256,balanced,0.05193600058555603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,256,balanced,0.057130664587020874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,256,balanced,0.08240533371766408
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,256,power_law_1.01,0.052928000688552856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,256,power_law_1.01,0.6177472114562989
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,6144,2048,8,256,1,256,power_law_1.2,3.885599899291992
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,256,power_law_1.01,0.05429760217666626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,256,power_law_1.01,0.8651904106140137
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,256,power_law_1.01,0.04287360012531281
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,256,power_law_1.01,0.04377599954605103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,256,power_law_1.01,1.1125375747680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,256,power_law_1.01,0.04438399970531463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,256,balanced,1.0383413632710774
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,256,power_law_1.01,0.04547840058803558
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,6144,2048,8,256,1,256,power_law_1.2,0.21313281059265138
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,256,power_law_1.01,0.04712960124015808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,256,power_law_1.01,1.5987327575683594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,256,balanced,0.1440000037352244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,256,balanced,1.4890559514363606
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,256,balanced,0.1460533340771993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,6144,2048,8,256,1,256,power_law_1.2,5.466041564941406
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,256,power_law_1.01,0.04776960015296936
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,6144,2048,8,256,1,256,power_law_1.2,0.02812800109386444
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,256,balanced,0.03596800069014231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,256,balanced,0.04942933221658071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,256,balanced,0.05064533154169718
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,6144,2048,8,256,1,256,power_law_1.2,0.027641600370407103
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,256,balanced,1.9397759437561035
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,6144,2048,8,256,1,256,power_law_1.2,0.02746880054473877
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,6144,2048,8,256,1,256,power_law_1.2,0.02967039942741394
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,256,balanced,2.8427254358927407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,6144,2048,8,256,1,256,power_law_1.2,0.031385600566864014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,6144,2048,8,256,1,256,power_law_1.2,0.07217919826507568
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,6144,2048,8,256,1,256,power_law_1.2,0.08819199800491333
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,256,balanced,0.14824000000953674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,6144,2048,8,256,1,256,power_law_1.2,0.12632319927215577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,256,balanced,0.15008533000946045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,256,balanced,0.15101333459218344
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,256,power_law_1.01,0.05063040256500244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,256,balanced,0.05644799768924713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,256,balanced,3.7455838521321616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,256,balanced,0.08581866820653279
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,6144,2048,8,256,1,256,power_law_1.2,0.22756481170654297
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,256,power_law_1.01,0.05841919779777527
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,256,power_law_1.01,2.0988351821899416
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,6144,2048,8,256,1,256,power_law_1.2,0.23185279369354247
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,6144,2048,8,256,1,256,power_law_1.2,0.23434879779815673
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,6144,2048,8,256,1,256,power_law_1.2,0.23850879669189454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,6144,2048,8,256,1,256,power_law_1.2,0.24560000896453857
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,6144,2048,8,256,1,256,power_law_1.2,0.24092159271240235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,256,power_law_1.01,3.098297691345215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,6144,2048,8,256,1,256,power_law_1.2,0.2491584062576294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,256,balanced,0.1564479966958364
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,6144,2048,8,256,1,256,power_law_1.2,0.2572927951812744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,256,balanced,0.16243732968966165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,256,balanced,0.1648319959640503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,256,balanced,0.14601066708564758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,256,balanced,0.17485332489013672
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,6144,2048,8,256,1,256,power_law_1.2,0.2637376070022583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,256,balanced,0.1848693291346232
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,256,power_law_1.01,0.06504960060119629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,256,balanced,0.19344000021616617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,6144,2048,8,256,1,256,power_law_1.2,0.2794624090194702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,256,balanced,0.21181867520014444
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,256,power_law_1.01,0.08740479946136474
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,256,balanced,0.25338133176167804
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,6144,2048,8,256,1,256,power_law_1.2,0.29700479507446287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,256,balanced,0.28998400767644245
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,256,power_law_1.01,0.10937600135803223
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,256,power_law_1.01,4.103699111938477
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,6144,2048,8,256,1,256,power_law_1.2,0.3113408088684082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,256,power_law_1.01,0.15351680517196656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,6144,2048,8,256,1,256,power_law_1.2,0.03994880020618439
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,6144,2048,8,256,1,256,power_law_1.2,0.34240639209747314
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,256,power_law_1.01,0.19410560131072999
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,6144,2048,8,256,1,256,power_law_1.2,0.31152639389038084
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,6144,2048,8,256,1,256,power_law_1.2,0.04115839898586273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,6144,2048,8,256,1,256,power_law_1.2,0.40215039253234863
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,256,power_law_1.01,0.27806720733642576
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,6144,2048,8,256,1,256,power_law_1.2,0.053472000360488894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,6144,2048,8,256,1,256,power_law_1.2,0.38052480220794677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,256,balanced,0.04749333361784617
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,6144,2048,8,256,1,256,power_law_1.2,0.46933760643005373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,256,balanced,0.14847466349601746
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,6144,2048,8,256,1,256,power_law_1.2,0.052288001775741576
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,256,balanced,0.1520960032939911
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,256,power_law_1.01,0.3704319953918457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,6144,2048,8,256,1,256,power_law_1.2,0.5123392105102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,256,balanced,0.1525813341140747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,256,balanced,0.15507733821868896
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,6144,2048,8,256,1,256,power_law_1.2,0.616864013671875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,256,balanced,0.1588159998257955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,256,balanced,0.1646773318449656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,256,power_law_1.01,0.5264895915985107
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,256,balanced,0.37060801188151044
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,6144,2048,8,256,1,256,power_law_1.2,0.77292799949646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,256,balanced,0.4555199940999349
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,256,balanced,0.6098026831944784
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,6144,2048,8,256,1,256,power_law_1.2,1.0470720291137696
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,256,power_law_1.01,0.6919040203094482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,256,balanced,0.7563947041829427
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,256,balanced,0.025237334271272022
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,256,power_law_1.01,0.030604800581932066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,256,balanced,1.0561119715372722
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,256,balanced,0.04704533517360687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,256,balanced,0.051514665285746254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,256,balanced,0.050623998045921326
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,6144,2048,8,256,1,256,power_law_1.2,1.369926357269287
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,256,balanced,0.05853333572546641
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,256,balanced,0.08897067109743755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,256,balanced,1.3517120679219563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,6144,2048,8,256,1,256,power_law_1.2,0.0536191999912262
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,256,balanced,0.09136000275611877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,256,balanced,0.09107733766237895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,256,balanced,0.09518399834632874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,256,balanced,0.170741339524587
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,6144,2048,8,256,1,256,power_law_1.2,0.042847999930381776
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,256,balanced,0.17917867501576742
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,256,balanced,0.18911999464035034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,256,balanced,0.19944000244140625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,256,balanced,0.2193653384844462
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,256,balanced,0.2625439961751302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,256,balanced,0.30080533027648926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,256,balanced,0.3845813274383545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,256,balanced,0.02532266577084859
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,256,balanced,0.025034666061401367
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,256,balanced,0.02719466636578242
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,256,balanced,0.02749866743882497
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,256,balanced,0.035391998787721
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,256,balanced,0.03572800010442734
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,256,balanced,0.037733333806196846
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,256,balanced,0.03782933453718821
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,6144,2048,8,256,1,256,power_law_1.2,0.6465472221374512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,256,balanced,0.039162665605545044
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,256,balanced,0.043791999419530235
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,256,balanced,0.03915733347336451
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,256,balanced,0.03839999934037527
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,256,balanced,0.03817066550254822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,6144,2048,8,256,1,256,power_law_1.2,0.9214591979980469
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,256,balanced,0.039706667264302574
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,256,balanced,0.03945599993069967
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,256,balanced,0.0418453315893809
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,256,balanced,0.04587733248869578
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,256,balanced,0.04799466828505198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,6144,2048,8,256,1,256,power_law_1.2,1.1808064460754395
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,256,balanced,0.4713759819666545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,256,balanced,0.6280320088068644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,256,power_law_1.01,0.05491200089454651
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,256,balanced,0.7863626480102539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,256,power_law_1.01,0.052102398872375486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,6144,2048,8,256,1,256,power_law_1.2,1.7319232940673828
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,256,balanced,0.09715732932090759
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,256,power_law_1.01,0.05573760271072388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,256,balanced,0.10331199566523235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,256,balanced,1.9477814038594563
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,6144,2048,8,256,1,256,power_law_1.2,0.043782401084899905
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,256,power_law_1.01,0.08046720027923585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,6144,2048,8,256,1,256,power_law_1.2,0.04411520063877106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,256,power_law_1.01,0.14203519821166993
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,6144,2048,8,256,1,256,power_law_1.2,2.2812736511230467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,6144,2048,8,256,1,256,power_law_1.2,0.0457152009010315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,256,power_law_1.01,0.14433280229568482
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,256,balanced,0.055888002117474876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,6144,2048,8,256,1,256,power_law_1.2,0.04737280011177063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,256,power_law_1.01,0.14606720209121704
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,6144,2048,8,256,1,256,power_law_1.2,0.04805760085582733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,256,power_law_1.01,0.14897919893264772
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,6144,2048,8,256,1,256,power_law_1.2,0.05219200253486633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,256,power_law_1.01,0.15008000135421753
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,256,balanced,1.094277302424113
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,256,balanced,0.10744000474611919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,256,power_law_1.01,0.1550655961036682
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,256,balanced,0.1092693308989207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,256,balanced,0.12170666456222534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,256,balanced,1.4047306378682454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,256,balanced,2.5361973444620767
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,256,balanced,2.0226613680521646
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,256,balanced,0.06235733131567637
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,256,balanced,0.08001066744327545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,256,balanced,0.09727467099825542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,256,balanced,0.1476693352063497
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,256,balanced,0.19139200448989868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,256,balanced,2.635642687479655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,256,balanced,0.2744266589482625
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,6144,2048,8,256,1,256,power_law_1.2,0.05960320234298706
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,256,balanced,0.3542506694793701
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,6144,2048,8,256,1,256,power_law_1.2,1.997337532043457
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,6144,2048,8,256,1,256,power_law_1.2,0.0690496027469635
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,256,balanced,0.13192533453305563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,256,balanced,0.14012799660364786
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,6144,2048,8,256,1,256,power_law_1.2,0.09066240191459655
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,256,balanced,0.15930666526158652
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,256,balanced,0.20174932479858398
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,6144,2048,8,256,1,256,power_law_1.2,0.11459200382232666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,256,balanced,0.24114133914311728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,256,balanced,0.32470399141311646
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,256,power_law_1.01,0.03351680040359497
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,6144,2048,8,256,1,256,power_law_1.2,0.16437760591506959
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,256,balanced,0.40651198228200275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,6144,2048,8,256,1,256,power_law_1.2,2.579955291748047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,256,power_law_1.01,0.04875519871711731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,256,balanced,0.5652000109354655
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,6144,2048,8,256,1,256,power_law_1.2,0.20906240940093995
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,256,power_law_1.01,0.048819199204444885
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,6144,2048,8,256,1,256,power_law_1.2,0.30789120197296144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,256,power_law_1.01,0.05581439733505249
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,6144,2048,8,256,1,256,power_law_1.2,3.344416046142578
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,256,power_law_1.01,0.04764159917831421
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,6144,2048,8,256,1,256,power_law_1.2,0.3987071990966797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,256,power_law_1.01,0.08401280045509338
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,256,power_law_1.01,0.15952639579772948
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,256,power_law_1.01,0.04803839921951294
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,256,power_law_1.01,0.14472320079803466
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,256,power_law_1.01,0.16430720090866088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,256,power_law_1.01,0.05022720098495483
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,256,power_law_1.01,0.14747519493103028
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,256,power_law_1.01,0.049907198548316954
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,256,power_law_1.01,0.02361599951982498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,256,power_law_1.01,0.14911359548568726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,256,power_law_1.01,0.05728639960289002
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,256,power_law_1.01,0.023398399353027344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,256,power_law_1.01,0.15198080539703368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,256,power_law_1.01,0.08673920035362244
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,6144,2048,8,256,1,256,power_law_1.2,4.554496002197266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,256,power_law_1.01,0.15367039442062377
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,256,power_law_1.01,0.08926720023155213
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,256,power_law_1.2,0.03219200074672699
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,256,power_law_1.01,0.15852799415588378
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,256,power_law_1.01,0.09079679846763611
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,256,power_law_1.2,0.05117440223693848
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,256,power_law_1.01,0.16405760049819945
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,256,power_law_1.01,0.0938431978225708
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,6144,2048,8,256,1,256,power_law_1.2,0.6118656158447265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,256,power_law_1.01,0.16826239824295045
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,256,power_law_1.01,0.09488000273704529
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,256,power_law_1.01,0.1786239981651306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,256,power_law_1.01,0.10021760463714599
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,6144,2048,8,256,1,256,power_law_1.2,0.7712959766387939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,256,power_law_1.01,0.18844799995422362
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,256,power_law_1.01,0.10531840324401856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,256,power_law_1.2,0.033504000306129454
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,256,balanced,0.7213599681854248
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,256,power_law_1.01,0.023705600202083586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,256,power_law_1.01,0.10902400016784668
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,256,power_law_1.2,0.0486272007226944
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,256,balanced,1.0286826292673747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,256,power_law_1.01,0.11987839937210083
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,256,power_law_1.2,0.05028480291366577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,256,power_law_1.01,0.12866560220718384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,256,balanced,1.336949348449707
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,256,power_law_1.2,0.0549888014793396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,256,power_law_1.01,0.13912320137023926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,256,power_law_1.2,0.052070397138595584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,256,power_law_1.2,0.08420479893684388
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,256,power_law_1.01,0.16075520515441893
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,256,power_law_1.01,0.17344640493392943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,256,balanced,1.9523305892944336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,256,power_law_1.01,0.20384640693664552
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,256,power_law_1.01,0.2432960033416748
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,256,power_law_1.01,0.19772160053253174
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,256,balanced,2.5648372968037925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,256,power_law_1.01,0.329638409614563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,256,power_law_1.01,0.21838080883026123
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,256,power_law_1.01,0.025382399559020996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,256,power_law_1.2,0.046963199973106384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,256,power_law_1.01,0.26153600215911865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,256,power_law_1.01,0.026387199759483337
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,256,power_law_1.2,0.04745599925518036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,256,power_law_1.01,0.30164480209350586
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,6144,2048,8,256,1,256,power_law_1.2,3.8916481018066404
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,256,power_law_1.01,0.03375999927520752
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,256,power_law_1.2,0.0499455988407135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,256,power_law_1.2,0.055641597509384154
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,256,power_law_1.01,0.03496319949626923
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,256,power_law_1.2,0.05041279792785645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,256,power_law_1.2,0.1444991946220398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,256,power_law_1.2,0.08142079710960388
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,256,power_law_1.01,0.03646079897880554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,256,power_law_1.2,0.05742719769477844
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,256,power_law_1.2,0.14762239456176757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,256,power_law_1.2,0.14161920547485352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,256,power_law_1.01,0.03692800104618073
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,256,power_law_1.2,0.0871295988559723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,256,power_law_1.2,0.14941439628601075
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,256,power_law_1.2,0.14458880424499512
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,256,power_law_1.01,0.03738879859447479
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,256,power_law_1.2,0.08915200233459472
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,256,power_law_1.2,0.15180799961090088
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,256,power_law_1.01,0.41754879951477053
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,256,power_law_1.01,0.04016000032424927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,256,power_law_1.2,0.09120000004768372
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,256,power_law_1.2,0.1537279963493347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,6144,2048,8,256,1,256,power_law_1.2,5.1318401336669925
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,256,power_law_1.01,0.5826560020446777
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,256,power_law_1.2,0.09400960206985473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,256,power_law_1.2,0.15845760107040405
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,256,power_law_1.01,0.390118408203125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,256,power_law_1.2,0.09544320106506347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,256,power_law_1.2,0.1641343951225281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,256,power_law_1.01,0.7425663948059082
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,256,power_law_1.01,0.18272000551223755
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,256,power_law_1.2,0.10065280199050904
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,256,power_law_1.01,0.47912960052490233
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,256,power_law_1.01,0.19290239810943605
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,256,power_law_1.2,0.10524159669876099
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,256,power_law_1.01,1.0700736045837402
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,256,power_law_1.01,0.6447807788848877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,256,power_law_1.2,0.10907520055770874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,256,power_law_1.2,0.12017279863357544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,256,power_law_1.01,0.8212544441223144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,256,power_law_1.2,0.14613120555877684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,256,power_law_1.2,0.13163520097732545
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,256,power_law_1.01,1.3859647750854491
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,256,power_law_1.01,0.03687680065631867
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,256,power_law_1.01,1.1566207885742188
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,256,power_law_1.01,0.036800000071525577
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,256,power_law_1.2,0.1678655982017517
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,256,power_law_1.01,2.021292877197266
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,256,power_law_1.01,1.4956735610961913
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,256,power_law_1.2,0.02640640139579773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,256,power_law_1.01,0.21119999885559082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,256,power_law_1.2,0.025171199440956117
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,256,power_law_1.2,0.025158399343490602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,256,power_law_1.01,2.682316780090332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,256,power_law_1.01,2.154412841796875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,256,power_law_1.2,0.1492095947265625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,256,balanced,0.03355200091997782
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,256,power_law_1.2,0.026694399118423463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,256,balanced,0.05266133447488149
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,256,balanced,0.05226666728655497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,256,balanced,0.05706666906674703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,256,power_law_1.2,0.15075839757919313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,256,balanced,0.08095466593901317
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,256,power_law_1.2,0.028217598795890808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,256,balanced,0.1421119968096415
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,256,balanced,0.14613866806030273
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,256,power_law_1.2,0.15549440383911134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,256,balanced,0.14816000064214072
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,256,power_law_1.2,0.03552640080451965
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,256,power_law_1.01,0.03661440014839172
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,256,balanced,0.15124266346295676
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,256,balanced,0.1523306667804718
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,256,balanced,0.15639999508857727
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,256,power_law_1.2,0.17769600152969361
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,256,power_law_1.2,0.03622399866580963
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,256,balanced,0.16075733304023743
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,256,power_law_1.01,2.854649543762207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,256,balanced,0.16665066281954447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,256,power_law_1.2,0.187225604057312
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,256,balanced,0.17490132649739584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,256,power_law_1.2,0.03776639997959137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,256,power_law_1.01,0.25252480506896974
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,256,power_law_1.2,0.19598079919815065
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,256,power_law_1.2,0.03793280124664307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,256,power_law_1.2,0.21692159175872802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,256,power_law_1.2,0.14111360311508178
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,256,power_law_1.2,0.038380798697471616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,256,power_law_1.2,0.25613439083099365
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,256,power_law_1.2,0.16440320014953613
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,256,power_law_1.2,0.0404992014169693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,256,power_law_1.2,0.29580800533294677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,256,power_law_1.2,0.20626559257507324
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,256,power_law_1.2,0.03842560052871704
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,256,power_law_1.2,0.38030080795288085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,256,power_law_1.2,0.1601408004760742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,256,power_law_1.2,0.038899201154708865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,256,power_law_1.01,0.03800320029258728
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,256,power_law_1.2,0.16459519863128663
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,256,power_law_1.2,0.03868800103664398
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,256,power_law_1.2,0.4727488040924072
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,256,balanced,0.03357866654793421
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,256,power_law_1.2,0.17443840503692626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,256,balanced,0.1844480037689209
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,256,power_law_1.2,0.03940480053424835
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,256,balanced,0.19339734315872192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,256,power_law_1.2,0.18426239490509033
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,256,power_law_1.2,0.6645440101623535
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,256,power_law_1.2,0.03935999870300293
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,256,power_law_1.01,0.2916224002838135
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,256,power_law_1.2,0.19325439929962157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,256,power_law_1.2,0.04196479916572571
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,256,power_law_1.01,0.37500159740447997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,256,power_law_1.2,0.8278464317321778
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,256,power_law_1.2,0.04524160027503967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,256,power_law_1.01,0.45763201713562013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,256,power_law_1.2,1.2007871627807618
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,256,power_law_1.2,0.2472127914428711
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,256,power_law_1.2,0.33911681175231934
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,256,power_law_1.01,0.03813120126724243
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,256,balanced,0.050399998823801674
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,256,balanced,0.05005866785844167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,256,power_law_1.2,1.583187198638916
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,256,balanced,0.057904000083605446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,256,power_law_1.01,0.0406464010477066
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,256,power_law_1.2,0.43084158897399905
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,256,power_law_1.01,0.04451839923858643
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,256,power_law_1.2,0.6072832107543945
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,256,power_law_1.01,0.046854400634765626
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,256,power_law_1.2,0.21319680213928222
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,256,power_law_1.01,0.055852800607681274
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,256,power_law_1.2,0.04847359955310822
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,256,power_law_1.2,2.3184768676757814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,256,power_law_1.2,0.7777984142303467
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,256,power_law_1.01,0.06607999801635742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,256,power_law_1.2,0.05891839861869812
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,256,power_law_1.01,0.6204864025115967
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,256,balanced,0.08470933636029561
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,256,balanced,0.14607999722162882
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,256,power_law_1.2,0.07033600211143494
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,256,power_law_1.01,0.08630399703979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,256,power_law_1.01,0.7755199909210205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,256,power_law_1.2,1.1080191612243653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,256,power_law_1.2,0.09157760143280029
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,256,power_law_1.2,3.046303939819336
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,256,power_law_1.2,0.11716480255126953
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,256,power_law_1.01,1.1084480285644531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,256,balanced,0.048986668388048805
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,256,balanced,0.04789866507053375
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,256,power_law_1.2,0.1755136013031006
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,256,power_law_1.2,0.23129599094390868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,256,power_law_1.01,1.4267775535583496
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,256,power_law_1.2,0.3364351987838745
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,256,balanced,0.21158399184544882
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,256,balanced,0.25222400824228924
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,256,balanced,0.1500320037206014
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,256,power_law_1.2,0.43824000358581544
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,256,balanced,0.15005866686503092
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,256,balanced,0.15246400237083435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,256,power_law_1.01,2.078579139709473
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,256,balanced,0.15448000033696493
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,256,balanced,0.1602666676044464
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,256,balanced,0.1667893330256144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,256,power_law_1.2,0.2542272090911865
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,256,balanced,0.16876266400019327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,256,balanced,0.17920533816019693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,256,balanced,0.19088000059127808
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,256,power_law_1.2,0.29589118957519533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,256,balanced,0.19779733816782633
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,256,balanced,0.21871467431386313
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,256,balanced,0.26079465945561725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,256,power_law_1.2,0.3809920072555542
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,256,balanced,0.30164267619450885
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,256,balanced,0.38610132535298664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,256,power_law_1.2,0.47019519805908205
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,256,balanced,0.4718879858652751
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,256,balanced,0.29205866654713947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,256,power_law_1.2,0.6622015953063964
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,256,balanced,0.37160531679789227
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,256,power_law_1.01,0.11218559741973877
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,256,balanced,0.4535040060679118
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,256,power_law_1.2,0.8324095726013183
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,256,balanced,0.6077226797739664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,256,power_law_1.2,1.4379776000976563
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,256,balanced,0.7585493723551432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,256,balanced,0.02720000098148982
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,256,balanced,0.027189334233601887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,256,balanced,0.02716800073782603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,256,power_law_1.2,1.186905574798584
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,256,balanced,0.02917333443959554
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,256,balanced,1.0571946303049724
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,256,balanced,0.02951466788848241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,256,power_law_1.01,2.7300735473632813
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,256,balanced,0.03559466699759165
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,256,balanced,0.03751999884843826
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,256,power_law_1.01,0.03057279884815216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,256,balanced,1.3523893356323242
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,256,power_law_1.2,2.161747169494629
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,256,power_law_1.2,1.525011157989502
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,256,power_law_1.01,0.052223998308181765
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,256,balanced,0.6297546625137329
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,256,balanced,1.9480427106221516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,256,balanced,0.7850399812062582
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,256,power_law_1.01,0.16232320070266723
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,256,balanced,1.0934826532999675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,256,balanced,0.052570665876070656
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,256,power_law_1.01,0.2124095916748047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,256,power_law_1.2,2.8738752365112306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,256,balanced,2.5416266123453775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,256,balanced,1.4029332796732585
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,256,power_law_1.01,0.3004863977432251
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,256,power_law_1.01,0.04593920111656189
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,256,balanced,0.03957866628964742
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,256,power_law_1.01,0.3916095972061157
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,256,balanced,0.03972800076007843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,256,balanced,0.03965866565704346
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,256,balanced,0.04387199878692627
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,256,balanced,0.03962666789690653
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,256,balanced,0.04043733328580856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,256,power_law_1.01,0.050521600246429446
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,256,power_law_1.01,0.05674880146980286
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,256,power_law_1.2,2.2413824081420897
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,256,balanced,0.05199466645717621
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,256,balanced,0.058677335580190025
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,256,balanced,0.08700799942016602
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,256,balanced,0.09103999535242717
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,256,power_law_1.01,0.034790399670600894
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,256,power_law_1.01,0.04971520006656647
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,256,power_law_1.01,0.04834559857845307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,256,balanced,2.0166932741800943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,256,power_law_1.01,0.05006719827651977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,256,power_law_1.01,0.04948480129241943
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,256,power_law_1.2,2.9970304489135744
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,256,balanced,0.03998400022586187
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,3072,1536,8,256,1,256,power_law_1.2,0.031839999556541446
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,256,balanced,0.041290665666262306
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,256,balanced,0.04159466673930486
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,256,balanced,2.6278613408406577
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,256,balanced,0.04359466830889384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,3072,1536,8,256,1,256,power_law_1.2,0.051744002103805545
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,256,balanced,0.04596266647179922
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,256,balanced,0.04805333415667216
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,256,power_law_1.01,0.08113279938697815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,3072,1536,8,256,1,256,power_law_1.2,0.05139840245246887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,256,balanced,0.05589866638183594
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,256,balanced,0.06229866544405619
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,256,balanced,0.0804799993832906
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,3072,1536,8,256,1,256,power_law_1.2,0.054099202156066895
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,256,balanced,0.09749866525332133
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,256,balanced,0.14798933267593384
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,3072,1536,8,256,1,256,power_law_1.2,0.08104320168495179
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,256,balanced,0.19310933351516724
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,256,balanced,0.27321066459019977
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,256,power_law_1.2,0.14156160354614258
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,256,balanced,0.3538399934768677
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,256,power_law_1.2,0.1446272015571594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,3072,1536,8,256,1,256,power_law_1.2,0.04718720018863678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,256,power_law_1.01,0.04979200065135956
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,256,power_law_1.2,0.1460096001625061
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,3072,1536,8,256,1,256,power_law_1.2,0.047539201378822324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,256,power_law_1.01,0.05675519704818725
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,256,power_law_1.2,0.14906879663467407
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,256,power_law_1.01,0.02377600073814392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,3072,1536,8,256,1,256,power_law_1.2,0.05050240159034729
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,256,power_law_1.01,0.08689919710159302
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,256,power_law_1.2,0.15058560371398927
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,256,power_law_1.01,0.023014399409294128
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,3072,1536,8,256,1,256,power_law_1.2,0.050783997774124144
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,3072,1536,8,256,1,256,power_law_1.01,0.141593599319458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,256,power_law_1.01,0.08951680064201355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,256,power_law_1.2,0.15544960498809815
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,3072,1536,8,256,1,256,power_law_1.2,0.05671679973602295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,256,power_law_1.01,0.0237184002995491
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,3072,1536,8,256,1,256,power_law_1.01,0.14433280229568482
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,256,power_law_1.01,0.09080960154533387
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,256,power_law_1.2,0.1601855993270874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,3072,1536,8,256,1,256,power_law_1.2,0.08735359907150268
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,256,power_law_1.2,0.16456960439682006
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,3072,1536,8,256,1,256,power_law_1.2,0.08978559970855712
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,256,power_law_1.2,0.17499519586563111
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,256,power_law_1.01,0.0549888014793396
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,256,power_law_1.2,0.18444160223007203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,256,power_law_1.2,0.09162880182266235
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,256,power_law_1.01,0.08364800214767457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,256,power_law_1.2,0.19288320541381837
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,256,power_law_1.2,0.09415680170059204
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,3072,1536,8,256,1,256,power_law_1.2,0.03366400003433227
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,256,power_law_1.2,0.21249279975891114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,256,power_law_1.01,0.14526720046997071
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,256,power_law_1.2,0.09545599818229675
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,3072,1536,8,256,1,256,power_law_1.2,0.04888319969177246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,256,power_law_1.2,0.2538687944412231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,3072,1536,8,256,1,256,balanced,0.09325333436330159
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,256,power_law_1.2,0.10053759813308716
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,256,balanced,0.09504000345865886
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,256,balanced,0.09710933764775594
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,256,power_law_1.2,0.1054144024848938
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,256,balanced,0.10111467043558757
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,256,balanced,0.10718933741251628
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,256,power_law_1.01,0.025337600708007814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,256,power_law_1.2,0.10980479717254639
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,3072,1536,8,256,1,256,power_law_1.01,0.09359999895095825
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,256,power_law_1.2,0.12024960517883301
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,256,power_law_1.01,0.026464000344276428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,3072,1536,8,256,1,256,power_law_1.01,0.09521920084953309
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,256,power_law_1.2,0.13100160360336305
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,3072,1536,8,256,1,256,power_law_1.01,0.10028799772262573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,256,power_law_1.2,0.13987200260162352
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,3072,1536,8,256,1,256,power_law_1.01,0.10528639554977418
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,256,power_law_1.2,0.1629631996154785
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,256,power_law_1.01,0.10931199789047241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,256,power_law_1.2,0.20663039684295653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,256,power_law_1.01,0.11909760236740112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,256,power_law_1.01,0.14802559614181518
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,3072,1536,8,256,1,256,power_law_1.2,0.04925439953804016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,256,power_law_1.2,0.24817919731140137
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,256,power_law_1.2,0.29594879150390624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,256,power_law_1.01,0.14985599517822265
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,3072,1536,8,256,1,256,power_law_1.2,0.05602560043334961
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,3072,1536,8,256,1,256,power_law_1.01,0.1463487982749939
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,256,power_law_1.2,0.3401727914810181
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,256,power_law_1.2,0.3803776025772095
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,3072,1536,8,256,1,256,power_law_1.2,0.08402559757232667
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,3072,1536,8,256,1,256,power_law_1.01,0.14877439737319947
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,256,power_law_1.2,0.4322239875793457
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,256,power_law_1.2,0.4663551807403564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,3072,1536,8,256,1,256,power_law_1.2,0.14446719884872436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,3072,1536,8,256,1,256,power_law_1.01,0.1503872036933899
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,256,power_law_1.01,0.033939200639724734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,256,power_law_1.2,0.6080063819885254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,3072,1536,8,256,1,256,power_law_1.01,0.15509120225906373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,256,power_law_1.2,0.6585087776184082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,256,power_law_1.01,0.03482879996299744
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,3072,1536,8,256,1,256,power_law_1.01,0.159334397315979
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,256,power_law_1.01,0.03616639971733093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,256,power_law_1.2,0.7678592205047607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,256,power_law_1.2,0.8347776412963868
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,256,power_law_1.01,0.03681280016899109
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,256,power_law_1.01,0.12859519720077514
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,256,power_law_1.01,0.03731839954853058
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,256,power_law_1.2,1.1210559844970702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,256,power_law_1.01,0.13923840522766112
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,256,power_law_1.2,1.2027456283569335
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,256,power_law_1.01,0.1525760054588318
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,256,power_law_1.01,0.04007680118083954
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,256,power_law_1.01,0.1601088047027588
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,256,power_law_1.01,0.154585599899292
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,256,power_law_1.01,0.03686400055885315
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,256,power_law_1.01,0.20295679569244385
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,256,power_law_1.2,1.4577856063842773
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,256,power_law_1.01,0.15915520191192628
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,256,power_law_1.2,1.5911487579345702
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,3072,1536,8,256,1,256,power_law_1.2,0.14740480184555055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,256,power_law_1.01,0.1650239944458008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,3072,1536,8,256,1,256,power_law_1.2,0.14883840084075928
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,256,power_law_1.01,0.1689919948577881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,3072,1536,8,256,1,256,power_law_1.01,0.1638975977897644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,3072,1536,8,256,1,256,power_law_1.2,0.15192960500717162
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,256,power_law_1.01,0.17978880405426026
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,3072,1536,8,256,1,256,power_law_1.01,0.17347840070724488
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,3072,1536,8,256,1,256,power_law_1.2,0.15370880365371703
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,256,power_law_1.01,0.18869760036468505
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,3072,1536,8,256,1,256,balanced,0.11063466469446818
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,3072,1536,8,256,1,256,balanced,0.12156266967455547
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,3072,1536,8,256,1,256,power_law_1.2,0.15846400260925292
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,3072,1536,8,256,1,256,balanced,0.12994133432706198
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,3072,1536,8,256,1,256,balanced,0.14029332995414734
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,3072,1536,8,256,1,256,power_law_1.2,0.1637760043144226
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,3072,1536,8,256,1,256,balanced,0.15954666336377463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,3072,1536,8,256,1,256,balanced,0.2014133334159851
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,3072,1536,8,256,1,256,power_law_1.2,0.16787199974060057
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,256,power_law_1.01,0.03694080114364624
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,256,power_law_1.01,0.24135680198669435
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,256,power_law_1.01,0.03660799860954285
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,256,power_law_1.01,0.33103361129760744
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,256,power_law_1.01,0.03800959885120392
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,256,power_law_1.01,0.03834240138530731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,256,power_law_1.01,0.4176959991455078
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,256,power_law_1.2,2.185215950012207
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,256,power_law_1.2,2.2899200439453127
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,256,power_law_1.01,0.04054400026798248
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,3072,1536,8,256,1,256,power_law_1.01,0.18267519474029542
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,256,power_law_1.01,0.04479359984397888
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,256,power_law_1.01,0.19886720180511475
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,3072,1536,8,256,1,256,power_law_1.01,0.19243520498275757
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,256,power_law_1.01,0.046911999583244324
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,256,power_law_1.01,0.21994879245758056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,3072,1536,8,256,1,256,power_law_1.01,0.21120638847351075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,256,power_law_1.01,0.055980801582336426
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,3072,1536,8,256,1,256,balanced,0.2424853245417277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,256,power_law_1.01,0.26220800876617434
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,3072,1536,8,256,1,256,power_law_1.01,0.25243520736694336
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,3072,1536,8,256,1,256,balanced,0.32342400153477985
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,256,power_law_1.01,0.06631680130958557
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,3072,1536,8,256,1,256,power_law_1.2,0.17816959619522094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,256,power_law_1.2,3.05218563079834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,3072,1536,8,256,1,256,balanced,0.4080746571222941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,256,power_law_1.01,0.30385279655456543
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,256,power_law_1.01,0.08773120045661927
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,256,balanced,0.5689493417739868
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,256,power_law_1.01,0.3907583951950073
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,256,power_law_1.01,0.11133439540863037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,256,balanced,0.7213226954142252
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,256,power_law_1.01,0.16173440217971802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,256,power_law_1.01,0.47956480979919436
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,3072,1536,8,256,1,256,power_law_1.01,0.585536003112793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,256,balanced,1.0296533107757568
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,256,power_law_1.01,0.20862081050872802
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,256,power_law_1.01,0.6465983867645264
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,3072,1536,8,256,1,256,power_law_1.01,0.7441088199615479
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,256,power_law_1.01,0.2995584011077881
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,256,power_law_1.2,2.879315185546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,256,balanced,0.04594666759173075
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,256,power_law_1.01,0.3900608062744141
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,256,balanced,0.0440533310174942
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,3072,1536,8,256,1,256,power_law_1.01,0.2919552087783813
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,3072,1536,8,256,1,256,power_law_1.01,1.0717056274414063
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,3072,1536,8,256,1,256,power_law_1.2,0.18892799615859984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,3072,1536,8,256,1,256,power_law_1.2,0.19747840166091918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,3072,1536,8,256,1,256,power_law_1.01,0.374726390838623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,3072,1536,8,256,1,256,power_law_1.2,0.218720006942749
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,256,power_law_1.01,1.3853055953979492
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,3072,1536,8,256,1,256,power_law_1.01,0.4593855857849121
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,3072,1536,8,256,1,256,power_law_1.2,0.2594815969467163
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1,3072,1536,8,256,1,256,power_law_1.2,0.02421119958162308
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,3072,1536,8,256,1,256,power_law_1.01,0.6181375980377197
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,3072,1536,8,256,1,256,power_law_1.2,0.30193281173706055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,3072,1536,8,256,1,256,balanced,1.338373343149821
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2,3072,1536,8,256,1,256,power_law_1.2,0.02436479926109314
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,256,power_law_1.01,2.039193534851074
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,3072,1536,8,256,1,256,power_law_1.2,0.3967616081237793
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,3072,1536,8,256,1,256,power_law_1.01,0.7848127841949463
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,256,balanced,0.04191466669241587
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,3072,1536,8,256,1,256,balanced,1.951632022857666
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,256,balanced,0.04386133452256521
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,3072,1536,8,256,1,256,power_law_1.2,0.4882175922393799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,3072,1536,8,256,1,256,power_law_1.01,1.1069375991821289
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,3072,1536,8,256,1,256,power_law_1.2,0.6736063957214355
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,256,power_law_1.01,2.6631807327270507
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,3072,1536,8,256,1,256,balanced,2.5634133021036782
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,256,balanced,0.06605866551399231
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,256,balanced,0.06769600013891856
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,3072,1536,8,256,1,256,power_law_1.01,1.4311231613159179
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,256,balanced,0.06635199983914693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,256,power_law_1.2,0.8567744255065918
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,256,balanced,0.10110400120417277
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,256,balanced,0.17062399784723917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,256,balanced,0.2344906727472941
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,256,balanced,0.24257065852483115
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4,3072,1536,8,256,1,256,power_law_1.2,0.0247871994972229
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,256,power_law_1.2,1.2188287734985352
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8,3072,1536,8,256,1,256,power_law_1.2,0.025676798820495606
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16,3072,1536,8,256,1,256,power_law_1.2,0.027385601401329042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,256,power_law_1.2,1.595359992980957
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,32,3072,1536,8,256,1,256,power_law_1.2,0.034143999218940735
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,48,3072,1536,8,256,1,256,power_law_1.2,0.0354559987783432
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,64,3072,1536,8,256,1,256,power_law_1.2,0.03682560026645661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,256,power_law_1.2,2.3184255599975585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,3072,1536,8,256,1,256,power_law_1.01,0.8219648361206054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,256,balanced,0.2526879906654358
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,80,3072,1536,8,256,1,256,power_law_1.2,0.03730559945106506
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,256,balanced,0.09074133634567261
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,256,balanced,0.2609226703643799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,256,balanced,0.16061333815256754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,256,balanced,0.2714453339576721
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,96,3072,1536,8,256,1,256,power_law_1.2,0.03809280097484589
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,256,balanced,0.09275733431180318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,256,balanced,0.28753066062927246
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,256,balanced,0.15847466389338175
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,256,balanced,0.30594666798909503
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,128,3072,1536,8,256,1,256,power_law_1.2,0.04126720130443573
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,256,power_law_1.01,0.07167360186576843
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,160,3072,1536,8,256,1,256,power_law_1.2,0.03829759955406189
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,256,power_law_1.2,3.130009651184082
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,192,3072,1536,8,256,1,256,power_law_1.2,0.03882879912853241
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,256,power_law_1.01,0.07271040081977845
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,256,3072,1536,8,256,1,256,power_law_1.2,0.039238399267196654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,256,power_law_1.01,0.0500544011592865
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,320,3072,1536,8,256,1,256,power_law_1.2,0.0392192006111145
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,256,power_law_1.01,0.05082240104675293
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,384,3072,1536,8,256,1,256,power_law_1.2,0.03957119882106781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,3072,1536,8,256,1,256,power_law_1.01,2.0883840560913085
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,512,3072,1536,8,256,1,256,power_law_1.2,0.04225279986858368
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,256,balanced,0.2903680006663005
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,3072,1536,8,256,1,256,power_law_1.01,1.1661184310913086
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,256,power_law_1.01,0.09185280203819275
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,256,balanced,0.2882346709569295
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,768,3072,1536,8,256,1,256,power_law_1.2,0.045817598700523376
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,256,balanced,0.32702932755152386
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1024,3072,1536,8,256,1,256,power_law_1.2,0.04991360008716583
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,256,balanced,0.36060798168182373
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,256,power_law_1.01,0.1625663995742798
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,256,balanced,0.396938681602478
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,3072,1536,8,256,1,256,power_law_1.01,1.5020671844482423
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,256,balanced,0.43535466988881427
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,256,balanced,0.5166079998016357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,256,balanced,0.6627253293991089
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,256,power_law_1.01,0.28682239055633546
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,256,balanced,0.8095680077870687
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,3072,1536,8,256,1,256,power_law_1.01,2.1810752868652346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,256,power_law_1.01,0.41011199951171873
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,256,balanced,0.4145119984944661
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,256,balanced,1.1068960030873616
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,256,balanced,0.42324801286061603
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,256,balanced,0.4138346513112386
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,256,balanced,0.4291253487269084
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,256,balanced,0.42258667945861816
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,256,balanced,1.3953812917073567
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,256,balanced,0.4368533293406169
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,256,balanced,0.43300799528757733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,256,balanced,0.44834665457407635
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,1536,3072,1536,8,256,1,256,power_law_1.2,0.06021760106086731
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,256,balanced,0.44090131918589276
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,256,balanced,0.46344534556070965
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,3072,1536,8,256,1,256,power_law_1.01,2.734732818603516
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,3072,1536,8,256,1,256,power_law_1.01,2.8883264541625975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,256,balanced,0.45001065731048584
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,256,balanced,0.48184001445770264
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,2048,3072,1536,8,256,1,256,power_law_1.2,0.07014399766921997
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,256,balanced,0.4691466490427653
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,256,balanced,0.5014506578445435
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,256,balanced,0.486410657564799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1,4096,2688,22,512,1,256,power_law_1.2,0.07121919989585876
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,3072,3072,1536,8,256,1,256,power_law_1.2,0.09353600144386291
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,256,balanced,0.5079573392868042
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,256,balanced,0.5414613485336304
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,4096,3072,1536,8,256,1,256,power_law_1.2,0.12289279699325562
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2,4096,2688,22,512,1,256,power_law_1.2,0.07157760262489318
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,256,balanced,0.5775839885075887
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,6144,3072,1536,8,256,1,256,power_law_1.2,0.173088002204895
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,256,balanced,0.6175520022710165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,256,power_law_1.01,0.4193535804748535
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,256,balanced,0.6987360318501791
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,256,balanced,0.8454773426055908
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,256,power_law_1.01,0.4291840076446533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,256,power_law_1.01,0.11658240556716919
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,256,power_law_1.01,0.43888001441955565
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,256,power_law_1.01,0.06754559874534607
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,256,power_law_1.01,0.15800960063934327
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,256,power_law_1.01,0.07279360294342041
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,256,power_law_1.01,0.2890624046325684
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4,4096,2688,22,512,1,256,power_law_1.2,0.11787519454956055
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,256,power_law_1.01,0.07242239713668823
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,8192,3072,1536,8,256,1,256,power_law_1.2,0.23079679012298585
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,256,power_law_1.01,0.411027193069458
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,256,power_law_1.01,0.09889919757843017
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,256,balanced,1.9818612734476726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,256,power_law_1.01,0.1673408031463623
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,256,balanced,0.5349013408025106
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,256,power_law_1.01,0.4486207962036133
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,256,balanced,0.5701706806818644
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,256,power_law_1.01,0.23300480842590332
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,256,balanced,2.552778720855713
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,256,balanced,0.6107733249664307
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,256,power_law_1.01,0.465664005279541
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,256,balanced,0.6845013300577799
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,256,power_law_1.01,0.24303359985351564
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,256,power_law_1.01,0.4827712059020996
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,256,balanced,0.8217546939849854
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,256,power_law_1.01,0.25107200145721437
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,256,balanced,0.9650239944458008
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8,4096,2688,22,512,1,256,power_law_1.2,0.15927040576934814
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,256,balanced,3.71836789449056
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,256,power_law_1.01,0.26012160778045657
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,256,balanced,0.9891093571980795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,256,balanced,1.246997356414795
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,256,power_law_1.01,0.2690367937088013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16,4096,2688,22,512,1,256,power_law_1.2,0.29069440364837645
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,256,balanced,1.2794666290283203
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,256,balanced,1.5240853627522786
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,256,power_law_1.01,0.286572790145874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,32,4096,2688,22,512,1,256,power_law_1.2,0.41276798248291013
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,256,balanced,1.5611039797465007
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,256,power_law_1.01,0.30392959117889407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,256,balanced,2.086496035257975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,256,power_law_1.2,0.421779203414917
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,256,balanced,2.132335980733236
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,256,power_law_1.01,0.503878402709961
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,12288,3072,1536,8,256,1,256,power_law_1.2,0.35433599948883054
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,256,balanced,2.6412320137023926
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,48,4096,2688,22,512,1,256,power_law_1.01,0.4196671962738037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,256,power_law_1.01,0.5380928039550781
VLLM,0.16.0,NVIDIA B200,moe,vllm_flashinfer_trtllm_moe_fp4,nvfp4,16384,3072,1536,8,256,1,256,power_law_1.2,0.4405375957489014
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,256,balanced,2.699082692464193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,256,power_law_1.01,0.4269440174102783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1,4096,2688,22,512,1,256,power_law_1.2,0.04745599925518036
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,256,power_law_1.01,0.4356095790863037
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,256,balanced,3.786522547403971
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2,4096,2688,22,512,1,256,power_law_1.2,0.049568000435829165
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,256,balanced,4.875397364298503
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,256,power_law_1.01,0.44417920112609866
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,256,power_law_1.01,0.3233664035797119
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,256,balanced,3.851776123046875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4,4096,2688,22,512,1,256,power_law_1.2,0.09230080246925354
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,256,power_law_1.01,0.3589695930480957
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,64,4096,2688,22,512,1,256,power_law_1.2,0.4299327850341797
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8,4096,2688,22,512,1,256,power_law_1.2,0.15690239667892455
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,256,power_law_1.01,0.40117759704589845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,80,4096,2688,22,512,1,256,power_law_1.2,0.43822078704833983
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,256,balanced,4.9286346435546875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,256,power_law_1.01,0.4428864002227783
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,256,power_law_1.01,0.5746816158294678
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,96,4096,2688,22,512,1,256,power_law_1.2,0.4462463855743408
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16,4096,2688,22,512,1,256,power_law_1.2,0.2873791933059692
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,256,power_law_1.01,0.5274943828582763
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,256,balanced,5.006682713826497
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,256,power_law_1.2,0.4633984088897705
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,256,power_law_1.01,0.6183872222900391
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,32,4096,2688,22,512,1,256,power_law_1.2,0.41118721961975097
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,256,balanced,7.22428830464681
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,256,power_law_1.01,0.6822591781616211
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,256,power_law_1.2,0.4806975841522217
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,256,power_law_1.01,0.6947135925292969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,48,4096,2688,22,512,1,256,power_law_1.2,0.4210368156433105
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,256,power_law_1.2,0.5009088039398193
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,128,4096,2688,22,512,1,256,power_law_1.01,0.46077442169189453
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,256,power_law_1.01,0.8392640113830566
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,256,power_law_1.01,0.8488191604614258
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,64,4096,2688,22,512,1,256,power_law_1.2,0.4304384231567383
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,256,power_law_1.2,0.5382336139678955
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,160,4096,2688,22,512,1,256,power_law_1.01,0.47777280807495115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,256,balanced,7.236565272013347
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,80,4096,2688,22,512,1,256,power_law_1.2,0.439404821395874
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,256,power_law_1.01,1.1468992233276367
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,256,power_law_1.01,1.0040127754211425
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,192,4096,2688,22,512,1,256,power_law_1.01,0.49826560020446775
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,256,power_law_1.2,0.5735360145568847
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,96,4096,2688,22,512,1,256,power_law_1.2,0.4478911876678467
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,256,balanced,7.366426467895508
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,256,4096,2688,22,512,1,256,power_law_1.01,0.5326720237731933
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,256,power_law_1.01,1.4455103874206543
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,256,power_law_1.2,0.6156864166259766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,320,4096,2688,22,512,1,256,power_law_1.01,0.5704959869384766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,256,power_law_1.2,0.6908095836639404
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,256,balanced,9.503621419270834
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,384,4096,2688,22,512,1,256,power_law_1.01,0.6106880187988282
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,256,power_law_1.01,2.0842048645019533
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,256,power_law_1.2,0.8425151824951171
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1,4096,2688,22,512,1,256,power_law_1.2,0.06854400038719177
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2,4096,2688,22,512,1,256,power_law_1.2,0.07219200134277344
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,256,power_law_1.2,1.0012672424316407
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,256,power_law_1.01,1.3148032188415528
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,128,4096,2688,22,512,1,256,power_law_1.2,0.46542720794677733
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4,4096,2688,22,512,1,256,power_law_1.2,0.07352960109710693
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,256,power_law_1.01,2.6788864135742188
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,256,balanced,9.67356808980306
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,160,4096,2688,22,512,1,256,power_law_1.2,0.4818880081176758
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8,4096,2688,22,512,1,256,power_law_1.2,0.10068479776382447
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,256,power_law_1.01,1.650009536743164
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,192,4096,2688,22,512,1,256,power_law_1.2,0.5027584075927735
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16,4096,2688,22,512,1,256,power_law_1.2,0.16706559658050538
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,512,4096,2688,22,512,1,256,power_law_1.01,0.6853119850158691
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,256,4096,2688,22,512,1,256,power_law_1.2,0.5395520210266114
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,32,4096,2688,22,512,1,256,power_law_1.2,0.23393280506134034
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,768,4096,2688,22,512,1,256,power_law_1.01,0.829139232635498
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,256,power_law_1.01,3.890860748291016
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,320,4096,2688,22,512,1,256,power_law_1.2,0.5738239765167237
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,256,balanced,9.505109151204428
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,48,4096,2688,22,512,1,256,power_law_1.2,0.24348158836364747
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1024,4096,2688,22,512,1,256,power_law_1.01,0.9798656463623047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,64,4096,2688,22,512,1,256,power_law_1.2,0.25102720260620115
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,256,power_law_1.2,1.3403648376464843
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,80,4096,2688,22,512,1,256,power_law_1.2,0.2607680082321167
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,96,4096,2688,22,512,1,256,power_law_1.2,0.2695872068405151
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,1536,4096,2688,22,512,1,256,power_law_1.01,1.2858943939208984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,128,4096,2688,22,512,1,256,power_law_1.2,0.28671998977661134
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,256,power_law_1.2,1.6679744720458984
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,256,power_law_1.01,2.267910385131836
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,2048,4096,2688,22,512,1,256,power_law_1.01,1.6132863998413085
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,256,power_law_1.01,5.104556655883789
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,384,4096,2688,22,512,1,256,power_law_1.2,0.615008020401001
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,256,power_law_1.2,2.3375871658325194
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,512,4096,2688,22,512,1,256,power_law_1.2,0.6948863983154296
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,256,power_law_1.01,2.921209526062012
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,3072,4096,2688,22,512,1,256,power_law_1.01,2.197056007385254
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,768,4096,2688,22,512,1,256,power_law_1.2,0.8589759826660156
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,160,4096,2688,22,512,1,256,power_law_1.2,0.3044800043106079
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1024,4096,2688,22,512,1,256,power_law_1.2,1.0154687881469726
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,256,power_law_1.2,2.9634559631347654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,4096,4096,2688,22,512,1,256,power_law_1.01,2.8133823394775392
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,256,power_law_1.01,4.2143104553222654
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,1536,4096,2688,22,512,1,256,power_law_1.2,1.374131202697754
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,256,power_law_1.01,7.5862274169921875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,2048,4096,2688,22,512,1,256,power_law_1.2,1.6986047744750976
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,256,power_law_1.2,4.322566223144531
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,6144,4096,2688,22,512,1,256,power_law_1.01,4.009587097167969
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,192,4096,2688,22,512,1,256,power_law_1.2,0.32527999877929686
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,3072,4096,2688,22,512,1,256,power_law_1.2,2.3075839996337892
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,256,4096,2688,22,512,1,256,power_law_1.2,0.3666752099990845
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,256,power_law_1.01,5.48023681640625
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,320,4096,2688,22,512,1,256,power_law_1.2,0.40340480804443357
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,384,4096,2688,22,512,1,256,power_law_1.2,0.44911999702453614
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,4096,4096,2688,22,512,1,256,power_law_1.2,2.9917055130004884
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,256,power_law_1.01,5.372582244873047
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,8192,4096,2688,22,512,1,256,power_law_1.2,5.914828872680664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,512,4096,2688,22,512,1,256,power_law_1.2,0.5333504199981689
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,768,4096,2688,22,512,1,256,power_law_1.2,0.6916351795196534
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,256,power_law_1.01,10.04031982421875
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1024,4096,2688,22,512,1,256,power_law_1.2,0.8533823966979981
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,6144,4096,2688,22,512,1,256,power_law_1.2,4.413580703735351
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,1536,4096,2688,22,512,1,256,power_law_1.2,1.1725312232971192
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,256,power_law_1.01,8.06495361328125
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,2048,4096,2688,22,512,1,256,power_law_1.2,1.4765695571899413
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,256,power_law_1.01,7.974803161621094
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,12288,4096,2688,22,512,1,256,power_law_1.2,9.110566711425781
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,3072,4096,2688,22,512,1,256,power_law_1.2,2.1079551696777346
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,8192,4096,2688,22,512,1,256,power_law_1.2,5.714579010009766
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,4096,4096,2688,22,512,1,256,power_law_1.2,2.7304895401000975
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,256,power_law_1.01,10.756883239746093
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,6144,4096,2688,22,512,1,256,power_law_1.2,4.094400024414062
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,256,power_law_1.01,10.704434967041015
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,8192,4096,2688,22,512,1,256,power_law_1.2,5.196780776977539
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,12288,4096,2688,22,512,1,256,power_law_1.2,7.9296318054199215
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,12288,4096,2688,22,512,1,256,power_law_1.2,8.752268981933593
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,float16,16384,4096,2688,22,512,1,256,power_law_1.2,12.062348937988281
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8_block,16384,4096,2688,22,512,1,256,power_law_1.2,10.70824966430664
VLLM,0.16.0,NVIDIA B200,moe,vllm_fused_moe,fp8,16384,4096,2688,22,512,1,256,power_law_1.2,11.751251220703125
